1 Introduction

The Open Data Protocol (OData) enables the creation of REST-based data services which allow resources, identified using Uniform Resource Locators (URLs) and defined in a data model, to be published and edited by Web clients using simple HTTP messages. This specification defines the core semantics and the behavioral aspects of the protocol.

The [OData‑URL] specification defines a set of rules for constructing URLs to identify the data and metadata exposed by an OData service as well as a set of reserved URL query options.

The [OData-CSDLJSON] specification defines a JSON representation of the entity data model exposed by an OData service.

The [OData-CSDLXML] specification defines an XML representation of the entity data model exposed by an OData service.

The [OData-JSON] document specifies the JSON format of the resource representations that are exchanged using OData.

1.0 IPR Policy

This specification is provided under the RF on RAND Terms Mode of the OASIS IPR Policy, the mode chosen when the Technical Committee was established. For information on whether any patents have been disclosed that may be essential to implementing this specification, and any offers of patent licensing terms, please refer to the Intellectual Property Rights section of the TC’s web page (https://www.oasis-open.org/committees/odata/ipr.php).

1.1 Terminology

The key words “MUST”, “MUST NOT”, “REQUIRED”, “SHALL”, “SHALL NOT”, “SHOULD”, “SHOULD NOT”, “RECOMMENDED”, “MAY”, and “OPTIONAL” in this document are to be interpreted as described in [RFC2119].

1.2 Normative References

[OData-ABNF] OData ABNF Construction Rules Version 4.01.
See link in "Additional artifacts" section on cover page.

[OData-Aggregation] OData Extension for Data Aggregation Version 4.0.
See link in "Related work" section on cover page.

[OData-CSDLJSON] OData Common Schema Definition Language (CSDL) JSON Representation Version 4.01. See link in "Related work" section on cover page.

[OData-CSDLXML] OData Common Schema Definition Language (CSDL) XML Representation Version 4.01. See link in "Related work" section on cover page

[OData-JSON] OData JSON Format Version 4.01.
See link in "Related work" section on cover page.

[OData-URL] OData Version 4.01 Part 2: URL Conventions.
See link in "Additional artifacts" section on cover page.

[OData-VocCap] OData Vocabularies Version 4.0: Capabilities Vocabulary.
See link in "Related work" section on cover page.

[OData-VocCore] OData Vocabularies Version 4.0: Core Vocabulary.
See link in "Related work" section on cover page.

[RFC2046] Freed, N. and N. Borenstein, "Multipurpose Internet Mail Extensions (MIME) Part Two: Media Types", RFC 2046, November 1996. https://tools.ietf.org/html/rfc2046.

[RFC2119] Bradner, S., “Key words for use in RFCs to Indicate Requirement Levels”, BCP 14, RFC 2119, March 1997. https://tools.ietf.org/html/rfc2119.

[RFC3987] Duerst, M. and, M. Suignard, “Internationalized Resource Identifiers (IRIs)”, RFC 3987, January 2005. https://tools.ietf.org/html/rfc3987.

[RFC5646] Phillips, A., Ed., and M. Davis, Ed., “Tags for Identifying Languages”, BCP 47, RFC 5646, September 2009. http://tools.ietf.org/html/rfc5646.

[RFC5789] Dusseault, L., and J. Snell, “Patch Method for HTTP”, RFC 5789, March 2010. http://tools.ietf.org/html/rfc5789.

[RFC7230] Fielding, R., Ed. and J. Reschke, Ed., “Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing”, RFC 7230, June 2014. https://tools.ietf.org/html/rfc7230.

[RFC7231] Fielding, R., Ed. and J. Reschke, Ed., “Hypertext Transfer Protocol (HTTP/1.1): Semantics and Content”, RFC 7231, June 2014. https://tools.ietf.org/html/rfc7231.

[RFC7232] Fielding, R., Ed. and J. Reschke, Ed., “Hypertext Transfer Protocol (HTTP/1.1): Conditional Requests”, RFC 7232, June 2014. https://tools.ietf.org/html/rfc7232.

[RFC7240] Snell, J., "Prefer Header for HTTP", RFC 7240, June 2014.https://tools.ietf.org/html/rfc7240.

[RFC7617] Reschke, J., "The 'Basic' HTTP Authentication Scheme", RFC 7617, September 2015. https://tools.ietf.org/html/rfc7617.

1.3 Typographical Conventions

Keywords defined by this specification use this monospaced font.

Normative source code uses this paragraph style.

Some sections of this specification are illustrated with non-normative examples.

Example 1: text describing an example uses this paragraph style

Non-normative examples use this paragraph style.

All examples in this document are non-normative and informative only.

All other text is normative unless otherwise labeled.

2 Overview

The OData Protocol is an application-level protocol for interacting with data via RESTful interfaces. The protocol supports the description of data models and the editing and querying of data according to those models. It provides facilities for:

· Metadata: a machine-readable description of the data model exposed by a particular service.

· Data: sets of data entities and the relationships between them.

· Querying: requesting that the service perform a set of filtering and other transformations to its data, then return the results.

· Editing: creating, updating, and deleting data.

· Operations: invoking custom logic

· Vocabularies: attaching custom semantics

The OData Protocol is different from other REST-based web service approaches in that it provides a uniform way to describe both the data and the data model. This improves semantic interoperability between systems and allows an ecosystem to emerge.

Towards that end, the OData Protocol follows these design principles:

· Prefer mechanisms that work on a variety of data sources. In particular, do not assume a relational data model.

· Extensibility is important. Services should be able to support extended functionality without breaking clients unaware of those extensions.

· Follow REST principles.

· OData should build incrementally. A very basic, compliant service should be easy to build, with additional work necessary only to support additional capabilities.

· Keep it simple. Address the common cases and provide extensibility where necessary.

3 Data Model

This section provides a high-level description of the Entity Data Model (EDM): the abstract data model that is used to describe the data exposed by an OData service. An OData Metadata Document is a representation of a service's data model exposed for client consumption.

The central concepts in the EDM are entities, relationships, entity sets, actions, and functions.

Entities are instances of entity types (e.g. Customer, Employee, etc.).

Entity types are named structured types with a key. They define the named properties and relationships of an entity. Entity types may derive by single inheritance from other entity types.

The key of an entity type is formed from a subset of the primitive properties (e.g. CustomerId, OrderId, LineId, etc.) of the entity type.

Complex types are keyless named structured types consisting of a set of properties. These are value types whose instances cannot be referenced outside of their containing entity. Complex types are commonly used as property values in an entity or as parameters to operations.

Properties declared as part of a structured type's definition are called declared properties. Instances of structured types may contain additional undeclared dynamic properties. A dynamic property cannot have the same name as a declared property. Entity or complex types which allow clients to persist additional undeclared properties are called open types.

Relationships from one entity to another are represented as navigation properties. Navigation properties are generally defined as part of an entity type, but can also appear on entity instances as undeclared dynamic navigation properties. Each relationship has a cardinality.

Enumeration types are named primitive types whose values are named constants with underlying integer values.

Type definitions are named primitive types with fixed facet values such as maximum length or precision. Type definitions can be used in place of primitive typed properties, for example, within property definitions.

Entity sets are named collections of entities (e.g. Customers is an entity set containing Customer entities). An entity's key uniquely identifies the entity within an entity set. If multiple entity sets use the same entity type, the same combination of key values can appear in more than one entity set and identifies different entities, one per entity set where this key combination appears. Each of these entities has a different entity-id. Entity sets provide entry points into the data model.

Operations allow the execution of custom logic on parts of a data model. Functions are operations that do not have side effects and may support further composition, for example, with additional filter operations, functions or an action. Actions are operations that allow side effects, such as data modification, and cannot be further composed in order to avoid non-deterministic behavior. Actions and functions are either bound to a type, enabling them to be called as members of an instance of that type, or unbound, in which case they are called as static operations. Action imports and function imports enable unbound actions and functions to be called from the service root.

Singletons are named entities which can be accessed as direct children of the entity container. A singleton may also be a member of an entity set.

An OData resource is anything in the model that can be addressed (an entity set, entity, property, or operation).

Refer to [OData-CSDLJSON] or [OData-CSDLXML] for more information on the OData entity data model.

3.1 Annotations

Model and instance elements can be decorated with Annotations.

Annotations can be used to specify an individual fact about an element, such as whether it is read-only, or to define a common concept, such as a person or a movie.

Applied annotations consist of a term (the namespace-qualified name of the annotation being applied), a target (the model or instance element to which the term is applied), and a value. The value may be a static value, or an expression that may contain a path to one or more properties of an annotated entity.

Annotation terms are defined in metadata and have a name and a type.

A set of related terms in a common namespace comprises a Vocabulary.

4 Service Model

OData services are defined using a common data model. The service advertises its concrete data model in a machine-readable form, allowing generic clients to interact with the service in a well-defined way.

An OData service exposes two well-defined resources that describe its data model; a service document and a metadata document.

The service document lists entity sets, functions, and singletons that can be retrieved. Clients can use the service document to navigate the model in a hypermedia-driven fashion.

The metadata document describes the types, sets, functions and actions understood by the OData service. Clients can use the metadata document to understand how to query and interact with entities in the service.

In addition to these two “fixed” resources, an OData service consists of dynamic resources. The URLs for many of these resources can be computed from the information in the metadata document.

See Requesting Data and Data Modification for details.

4.1 Entity-Ids and Entity References

Whereas entities within an entity set are uniquely identified by their key values, entities are also uniquely identified by a durable, opaque, globally unique entity-id. The entity-id MUST be an IRI as defined in [RFC3987] and MAY be expressed in payloads, headers, and URLs as a relative reference as appropriate. While the client MUST be prepared to accept any IRI, services MUST use valid URIs in this version of the specification since there is currently no lossless representation of an IRI in the EntityId header.

Services are strongly encouraged to use the canonical URL for an entity as defined in OData-URL as its entity-id, but clients cannot assume the entity-id can be used to locate the entity unless the Core.DereferenceableIDs term is applied to the entity container, nor can the client assume any semantics from the structure of the entity-id. The canonical resource $entity provides a general mechanism for resolving an entity-id into an entity representation.

Services that use the standard URL conventions for entity-ids annotate their entity container with the term Core.ConventionalIDs, see [OData-VocCore].

Entity references refer to an entity using the entity's entity-id.

4.2 Read URLs and Edit URLs

The read URL of an entity is the URL that can be used to read the entity.

The edit URL of an entity is the URL that can be used to update or delete the entity.

The edit URL of a property is the edit URL of the entity with appended segment(s) containing the path to the property.

Services are strongly encouraged to use the canonical URL for an entity as defined in OData-URL for both the read URL and the edit URL of an entity, with a cast segment to the type of the entity appended to the canonical URL if the type of the entity is derived from the declared type of the entity set. However, clients cannot assume this convention and must use the links specified in the payload according to the appropriate format as the two URLs may be different from one another, or one or both of them may differ from convention.

4.3 Transient Entities

Transient entities are instances of an entity type that are “calculated on the fly” and only exist within a single payload. They cannot be reread or updated and consequently possess neither a stable entity-id nor a read URL or an update URL.

4.4 Default Namespaces

References to actions, functions, and types within a URL typically requires prefixing the name of the action, function, or type with the namespace or alias in which it is defined. This namespace qualification enables differentiating between similarly named elements across schema, or between properties and bound functions, actions, or types with the same name.

Services MAY define one or more default namespaces through the Core.DefaultNamespace term defined in [OData-VocCore]. Functions, actions and types in a default namespace can be referenced in URLs with or without namespace or alias qualification.

Service designers should ensure uniqueness of schema children across all default namespaces, and should avoid naming bound functions, actions, or derived types with the same name as a structural or navigation property of the type.

In the case where ambiguity does exist, an unqualified segment appended to a structured value is always first compared to the list of properties defined on the structured type. If no defined property with a name matching the unqualified segment exists, or the preceding segment represents a collection or a scalar value, it is next compared to the names of any bound functions or actions, or derived type names, defined within any default namespace. If it still does not match, and the preceding segment represents a structured value, it is interpreted as a dynamic property.

Services MAY disallow dynamic properties on structured values whose names conflict with a bound action, function, or derived type defined within in a default namespace.

The behavior if name conflicts occur across children of default namespaces is undefined. Generic clients are encouraged to always qualify action, function, and type names in order to avoid any possible ambiguity.

5 Versioning

Versioning enables clients and services to evolve independently. OData defines semantics for both protocol and data model versioning.

5.1 Protocol Versioning

OData requests and responses are versioned according to the OData-Version header.

OData clients include the OData-MaxVersion header in requests in order to specify the maximum acceptable response version. Services respond with the maximum supported version that is less than or equal to the requested OData-MaxVersion, using decimal comparison. The syntax of the OData-Version and OData-MaxVersion header fields is defined in [OData-ABNF].

Services SHOULD advertise supported versions of OData through the Core.ODataVersions term, defined in [OData-VocCore].

This version of the specification defines OData version values 4.0 and 4.01. Content that applies only to one version or another is explicitly called out in the text.

5.2 Model Versioning

The Data Model exposed by an OData Service defines a contract between the OData service and its clients. Services are allowed to extend their model only to the degree that it does not break existing clients. Breaking changes, such as removing properties, changing the type of existing properties, adding or removing key properties, or reordering action or function parameters, require that a new service version is provided at a different service root URL for the new model, or that the service version its metadata using the Core.SchemaVersion annotation, defined in [OData-VocCore].

Services that version their metadata MUST support version-specific requests according to the $schemaversion system query option. The following Data Model additions are considered safe and do not require services to version their entry point or schema.

· Adding a property that is nullable or has a default value; if it has the same name as an existing dynamic property, it must have the same type (or base type) as the existing dynamic property

· Adding a navigation property that is nullable or collection-valued; if it has the same name as an existing dynamic navigation property, it must have the same type (or base type) as the existing dynamic navigation property

· Adding a new entity type to the model

· Adding a new complex type to the model

· Adding a new entity set

· Adding a new singleton

· Adding an action, a function, an action import, or function import

· Adding an action parameter that is nullable after existing parameters

· Adding an action or function parameter that is annotated with Core.OptionalParameter after existing parameters

· Adding a type definition or enumeration

· Adding a new term

· Adding any annotation to a model element that does not need to be understood by the client in order to correctly interact with the service

Clients SHOULD be prepared for services to make such incremental changes to their model. In particular, clients SHOULD be prepared to receive properties and derived types not previously defined by the service.

Services SHOULD NOT change their data model depending on the authenticated user. If the data model is user or user-group dependent, all changes MUST be safe changes as defined in this section when comparing the full model to the model visible to users with restricted authorizations.

6 Extensibility

The OData protocol supports both user- and version-driven extensibility through a combination of versioning, convention, and explicit extension points.

6.1 Query Option Extensibility

Query options within the request URL can control how a particular request is processed by the service.

OData-defined system query options are optionally prefixed with "$". Services may support additional custom query options not defined in the OData specification, but they MUST NOT begin with the "$" or "@" character and MUST NOT conflict with any OData-defined system query options defined in the OData version supported by the service.

OData services SHOULD NOT require any query options to be specified in a request. Services SHOULD fail any request that contains query options that they do not understand and MUST fail any request that contains unsupported OData query options defined in the version of this specification supported by the service.

In many cases OData services return URLs to identify resources that are later requested by clients. Where possible, interoperability is enhanced by providing all identifying information in the path portion of the URL. However, clients should be prepared for such URLs to include custom query options and propagate any such custom query options in future requests to the identified resource.

6.2 Payload Extensibility

OData supports extensibility in the payload, according to the specific format.

Regardless of the format, additional content MUST NOT be present if it needs to be understood by the receiver in order to correctly interpret the payload according to the specified OData-Version header. Thus, clients and services MUST be prepared to handle or safely ignore any content not specifically defined in the version of the payload specified by the OData-Version header.

6.3 Action/Function Extensibility

Actions and Functions extend the set of operations that can be performed on or with a service or resource. Actions can have side-effects. For example, Actions can be used to modify data or to invoke custom operations. Functions MUST NOT have side-effects. Functions can be invoked from a URL that addresses a resource or within an expression to a $filter or $orderby system query option.

Fully qualified action and function names include a namespace or alias prefix. The Edm, odata and geo namespaces are reserved for the use of this specification.

An OData service MUST fail any request that contains actions or functions that it does not understand.

6.4 Vocabulary Extensibility

The set of annotations defined within a schema comprise a vocabulary. Shared vocabularies provide a powerful extensibility point for OData.

Metadata annotations can be used to define additional characteristics or capabilities of a metadata element, such as a service, entity type, property, function, action or parameter. For example, a metadata annotation could define ranges of valid values for a particular property.

Instance annotations can be used to define additional information associated with a particular result, entity, property, or error; for example whether a property is read-only for a particular instance.

Where annotations apply across all instances of a type, services are encouraged to specify the annotation in metadata rather than repeating in each instance of the payload. Where the same annotation is defined at both the metadata and instance level, the instance-level annotation overrides the one specified at the metadata level.

A service MUST NOT require the client to understand custom annotations in order to accurately interpret a response.

OData defines a Core vocabulary with a set of basic terms describing behavioral aspects along with terms that can be used in defining other vocabularies; see [OData-VocCore].

6.5 Header Field Extensibility

OData defines semantics around certain HTTP request and response headers. Services that support a version of OData conform to the processing requirements for the headers defined by this specification for that version.

Individual services may define custom headers. These headers MUST NOT begin with OData. Custom headers SHOULD be optional when making requests to the service. A service MUST NOT require the client to understand custom headers to accurately interpret the response.

6.6 Format Extensibility

An OData service MUST support [OData-JSON] and MAY support additional formats for both request and response bodies.

7 Formats

The client MAY request a particular response format through the Accept header, as defined in [RFC7231], or through the system query option $format.

In the case that both the Accept header and the $format system query option are specified on a request, the value specified in the $format query option MUST be used.

If the service does not support the requested format, it replies with a 406 Not Acceptable error response.

Services SHOULD advertise their supported formats in the metadata document by annotating their entity container with the term Capabilities.SupportedFormats, as defined in [OData-VocCap], listing all available formats and combinations of supported format parameters.

The media types for the JSON and XML representation of the metadata document are described in section “Metadata Document Request”.

The format specification [OData-JSON] describes the media type and the format parameters for non-metadata requests and responses.

For non-metadata requests, if neither the Accept header nor the $format query option are specified, the service MAY respond to requests in any format.

Client libraries MUST retain the order of objects within an array in JSON responses, and elements in document order for XML responses, including CSDL documents.

8 Header Fields

OData defines semantics around the following request and response headers. Additional headers may be specified, but have no unique semantics defined in OData.

8.1 Common Headers

The following headers are common between OData requests and responses.

8.1.1 Header Content-Type

The format of a non-empty individual request or response body, alone or within a batch, MUST be specified in the Content-Type header of a request or response. The exception to this is if the body represents the media stream of a media entity or stream property, in which case the Content-Type header SHOULD be present.

The specified format MAY include format parameters. Clients MUST be prepared for the service to return custom format parameters not defined in OData and SHOULD NOT expect that such format parameters can be ignored. Custom format parameters MUST NOT start with "odata" and services MUST NOT require generic OData consumers to understand custom format parameters in order to correctly interpret the payload.

See [OData-JSON] for format-specific details about format parameters within the Content-Type header.

8.1.2 Header Content-Encoding

As defined in [RFC7231], the Content-Encoding header field is used as a modifier to the media-type (as indicated in the Content-Type). When present, its value indicates what additional content codings have been applied to the entity-body.
A service MAY specify a list of acceptable content codings using an annotation with term Capabilities.AcceptableEncodings, see [OData-VocCap].

If the Content-Encoding header is specified on an individual request or response within a batch, then it specifies the encoding for that individual request or response. Individual requests or responses that don’t include the Content-Encoding header inherit the encoding of the overall batch request or response.

8.1.3 Header Content-Language

As defined in [RFC7231], a request or response can include a Content-Language header to indicate the natural language of the intended audience for the enclosed message body. OData does not add any additional requirements over HTTP for including Content-Language. OData services can annotate model elements whose content depends on the content language with the term Core.IsLanguageDependent, see [OData-VocCore].

If the Content-Language header is specified on an individual request or response within a batch, then it specifies the language for that individual request or response. Individual requests or responses that don’t include the Content-Language header inherit the language of the overall batch request or response.

8.1.4 Header Content-Length

As defined in [RFC7230], a request or response SHOULD include a Content-Length header when the message's length can be determined prior to being transferred. OData does not add any additional requirements over HTTP for writing Content-Length.

If the Content-Length header is specified on an individual request or response within a batch, then it specifies the length for that individual request or response.

8.1.5 Header OData-Version

OData clients SHOULD use the OData-Version header on a request to specify the version of the protocol used to generate the request payload.

If present on a request, the service MUST interpret the request payload according to the rules defined in the specified version of the protocol or fail the request with a 4xx response code.

If not specified in a request, the service MUST assume the request payload is generated using the minimum of the OData-MaxVersion, if specified, and the maximum version of the protocol that the service understands.

OData services MUST include the OData-Version header on a response to specify the version of the protocol used to generate the response payload. The client MUST interpret the response payload according to the rules defined in the specified version of the protocol. Request and response payloads are independent and may have different OData-Version headers according to the above rules.

For more details, see Versioning.

If the OData-Version header is specified on an individual request or response within a batch, then it specifies the OData version for that individual request or response. Individual requests or responses that don’t include the OData-Version header inherit the OData version of the overall batch request or response. This OData version does not typically vary within a batch.

8.2 Request Headers

In addition to the Common Headers, the client may specify any combination of the following request headers.

8.2.1 Header Accept

As defined in [RFC7231], the client MAY specify the set of accepted formats with the Accept Header.

Services MUST reject formats that specify unknown or unsupported format parameters.

If a media type specified in the Accept header includes a charset format parameter and the request also contains an Accept-Charset header, then the Accept-Charset header MUST be used.

If the media type specified in the Accept header does not include a charset format parameter, then the Content-Type header of the response MUST NOT contain a charset format parameter.

The service SHOULD NOT add any format parameters to the Content-Type parameter not specified in the Accept header.

If the Accept header is specified on an individual request within a batch, then it specifies the acceptable formats for that individual request. Requests within a batch that don’t include the Accept header inherit the acceptable formats of the overall batch request.

8.2.2 Header Accept-Charset

As defined in [RFC7231], the client MAY specify the set of accepted character sets with the Accept-Charset header.

If the Accept-Charset header is specified on an individual request within a batch, then it specifies the acceptable character sets for that individual request. Requests within a batch that don’t include the Accept-Charset header inherit the acceptable character sets of the overall batch request.

8.2.3 Header Accept-Language

As defined in [RFC7231], the client MAY specify the set of accepted natural languages with the Accept-Language header.

If the Accept-Language header is specified on an individual request within a batch, then it specifies the acceptable languages for that individual request. Requests within a batch that don’t include the Accept-Language header inherit the acceptable languages of the overall batch request.

8.2.4 Header If-Match

As defined in [RFC7232], a client MAY include an If-Match header in a request to GET, POST, PUT, PATCH or DELETE. The value of the If-Match request header MUST be an ETag value previously retrieved for the resource, or * to match any value.

If an operation on an existing resource requires an ETag, (see term Core.OptimisticConcurrency in [OData-VocCore] and property OptimisticConcurrencyControl of type Capabilities.NavigationPropertyRestriction in [OData-VocCap]) and the client does not specify an If-Match request header in a Data Modification Request or in an Action Request invoking an action bound to the resource, the service responds with a 428 Precondition Required and MUST ensure that no observable change occurs as a result of the request.

If present, the request MUST only be processed if the specified ETag value matches the current ETag value of the target resource. Services sending ETag headers with weak ETags that only depend on the representation-independent entity state MUST use the weak comparison function because it is sufficient to prevent accidental overwrites. This is a deviation from [RFC7232].

If the value does not match the current ETag value of the resource for a Data Modification Request or Action Request, the service MUST respond with 412 Precondition Failed and MUST ensure that no observable change occurs as a result of the request. In the case of an upsert, if the addressed entity does not exist the provided ETag value is considered not to match.

An If-Match header with a value of * in a PUT or PATCH request results in an upsert request being processed as an update and not an insert.

The If-Match header MUST NOT be specified on a batch request, but MAY be specified on individual requests within the batch.

8.2.5 Header If-None-Match

As defined in [RFC7232], a client MAY include an If-None-Match header in a request to GET, POST, PUT, PATCH or DELETE. The value of the If-None-Match request header MUST be an ETag value previously retrieved for the resource, or *.

If present, the request MUST only be processed if the specified ETag value does not match the current ETag value of the resource, using the weak comparison function (see [RFC7232]). If the value matches the current ETag value of the resource, then for a GET request, the service SHOULD respond with 304 Not Modified, and for a Data Modification Request or Action Request, the service MUST respond with 412 Precondition Failed and MUST ensure that no observable change occurs as a result of the request.

An If-None-Match header with a value of * in a PUT or PATCH request results in an upsert request being processed as an insert and not an update.

The If-None-Match header MUST NOT be specified on a batch request, but MAY be specified on individual requests within the batch.

8.2.6 Header Isolation (OData-Isolation)

The Isolation header specifies the isolation of the current request from external changes. The only supported value for this header is snapshot.

If the service doesn’t support Isolation:snapshot and this header was specified on the request, the service MUST NOT process the request and MUST respond with 412 Precondition Failed.

Snapshot isolation guarantees that all data returned for a request, including multiple requests within a batch or results retrieved across multiple pages, will be consistent as of a single point in time. Only data modifications made within the request (for example, by a data modification request within the same batch) are visible. The effect is as if the request generates a "snapshot" of the committed data as it existed at the start of the request.

The Isolation header may be specified on a single or batch request. If it is specified on a batch then the value is applied to all statements within the batch.

Next links returned within a snapshot return results within the same snapshot as the initial request; the client is not required to repeat the header on each individual page request.

The Isolation header has no effect on links other than the next link. Navigation links, read links, and edit links return the current version of the data.

A service returns 410 Gone or 404 Not Found if a consumer tries to follow a next link referring to a snapshot that is no longer available.

The syntax of the Isolation header is defined in [OData-ABNF].

A service MAY specify the support for Isolation:snapshot using an annotation with term Capabilities.IsolationSupported, see [OData-VocCap].

Note: The Isolation header was named OData-Isolation in OData version 4.0. Services that support the Isolation header SHOULD also support OData-Isolation for OData 4.0 clients and clients SHOULD use OData-Isolation for compatibility with OData 4.0 services. If both Isolation and OData-Isolation headers are specified in the same request, the value of the Isolation preference SHOULD be used.

8.2.7 Header OData-MaxVersion

Clients SHOULD specify an OData-MaxVersion request header.

If specified, the service MUST generate a response with the greatest supported OData-Version less than or equal to the specified OData-MaxVersion.

If OData-MaxVersion is not specified, then the service SHOULD return responses with the same OData version over time and interpret the request as having an OData-MaxVersion equal to the maximum OData version supported by the service at its initial publication.

If the OData-MaxVersion header is specified on an individual request within a batch, then it specifies the maximum OData version for that individual request. Individual requests that don’t include the OData-MaxVersion header inherit the maximum OData version of the overall batch request or response. The maximum OData version does not typically vary within a batch.

For more details, see Versioning.

8.2.8 Header Prefer

The Prefer header, as defined in [RFC7240], allows clients to request certain behavior from the service. The service MUST ignore preference values that are either not supported or not known by the service.

The value of the Prefer header is a comma-separated list of preferences. The following subsections describe preferences whose meaning in OData is defined by this specification.

In response to a request containing a Prefer header, the service MAY return the Preference-Applied and Vary headers.

8.2.8.1 Preference allow-entityreferences (odata.allow-entityreferences)

The allow-entityreferences preference indicates that the service is allowed to return entity references in place of entities that have previously been returned, with at least the properties requested, in the same response (for example, when serializing the expanded results of many-to-many relationships). The service MUST NOT return entity references in place of requested entities if allow-entityreferences has not been specified in the request, unless explicitly defined by other rules in this document. The syntax of the allow-entityreferences preference is defined in [OData-ABNF].

In the case the service applies the allow-entityreferences preference it MUST include a Preference-Applied response header containing the allow-entityreferences preference to indicate that entity references MAY be returned in place of entities that have previously been returned.

If the allow-entityreferences preference is specified on an individual request within a batch, then it specifies the preference for that individual request. Individual requests within a batch that don’t include the allow-entityreferences preference inherit the preference of the overall batch request.

Note: The allow-entityreferences preference was named odata.allow-entityreferences in OData version 4.0. Services that support the allow-entityreferences preference SHOULD also support odata.allow-entityreferences for OData 4.0 clients and clients SHOULD use odata.allow-entityreferences for compatibility with OData 4.0 services.

8.2.8.2 Preference callback (odata.callback)

For scenarios in which links returned by the service are used by the client to poll for additional information, the client can specify the callback preference to request that the service notify the client when data is available.

The callback preference can be specified:

when requesting asynchronous processing of a request with the respond-async preference, or
on a GET request to a delta link.

The callback preference MUST include the parameter url whose value is the URL of a callback endpoint to be invoked by the OData service when data is available. The syntax of the callback preference is defined in [OData-ABNF].

For HTTP based callbacks, the OData service executes an HTTP GET request against the specified URL.

Services that support callback SHOULD support notifying the client through HTTP. Services can advertise callback support using the Capabilities.CallbackSupported annotation term defined in [OData-VocCap].

If the service applies the callback preference it MUST include the callback preference in the Preference-Applied response header.

When the callback preference is applied to asynchronous requests, the OData service invokes the callback endpoint once it has finished processing the request. The status monitor resource, returned in the Location header of the previously returned 202 Accepted response, can then be used to retrieve the results of the asynchronously executed request.

When the callback preference is specified on a GET request to a delta link and there are no changes available, the OData service returns a 202 Accepted response with a Location header specifying the delta link to be used to check for future updates. The OData service then invokes the specified callback endpoint once new changes become available.

Combining respond-async, callback and track-changes preferences on a GET request to a delta-link might influence the response in a couple of ways.

· If the service processes the request synchronously, and no updates are available, then the response is the same as if the respond-async hadn’t been specified and results in a response as described above.

If the service processes the request asynchronously, then it responds with a 202 Accepted response specifying the URL to the status monitor resource as it would have with any other asynchronous request. Once the service has finished processing the asynchronous request to the delta link resource, if changes are available it invokes the specified callback endpoint. If no changes are available, the service SHOULD wait to notify the client until changes are available. Once notified, the client uses the status monitor resource from the Location header of the previously returned 202 Accepted response to retrieve the results. In case no updates were available after processing the initial request, the result will contain no updates and the client can use the delta-link contained in the result to retrieve the updates that have since become available.

If the consumer specifies the same URL as callback endpoint in multiple requests, the service MAY collate them into a single notification once additional data is available for any of the requests. However, the consumer MUST be prepared to deal with receiving up to as many notifications as it requested.

Example 2: using a HTTP callback endpoint to receive notification

Prefer: callback; url="http://myserver/notfication/token/12345"

If the callback preference is specified on an individual request within a batch, then it specifies the callback to be used for tracking changes to that individual request. If the callback preference is specified on a batch, then it specifies the callback to be used for async responses to the batch.

Note: The callback preference was named odata.callback in OData version 4.0. Services that support the callback preference SHOULD also support odata.callback for OData 4.0 clients and clients SHOULD use odata.callback for compatibility with OData 4.0 services. If both callback and odata.callback preferences are specified in the same request, the value of the callback preference SHOULD be used.

8.2.8.3 Preference continue-on-error (odata.continue-on-error)

The continue-on-error preference on a batch request is used to request whether, upon encountering a request within the batch that returns an error, the service return the error for that request and continue processing additional requests within the batch (if specified with an implicit or explicit value of true), or rather stop further processing (if specified with an explicit value of false). The syntax of the continue-on-error preference is defined in [OData-ABNF].

The continue-on-error preference can also be used on a delta update, set-based update, or set-based delete to request that the service continue attempting to process changes after receiving an error.

A service MAY specify support for the continue-on-error preference using an annotation with term Capabilities.BatchContinueOnErrorSupported, see [OData-VocCap].

The continue-on-error preference SHOULD NOT be applied to individual requests within a batch.

Note: The continue-on-error preference was named odata.continue-on-error in OData version 4.0. Services that support the continue-on-error preference SHOULD also support odata.continue-on-error for OData 4.0 clients and clients SHOULD use odata.continue-on-error for compatibility with OData 4.0 services.

8.2.8.4 Preference include-annotations (odata.include-annotations)

The include-annotations preference in a request for data or metadata is used to specify the set of annotations the client requests to be included, where applicable, in the response.

The value of the include-annotations preference is a comma-separated list of namespace-qualified term names or term name patterns to include or exclude, with * as a wildcard for name segments. Term names and term name patterns can optionally be followed by a hash (#) character and an annotation qualifier. The full syntax of the include-annotations preference is defined in [OData-ABNF].

The most specific identifier always takes precedence, with an explicit name taking precedence over a name pattern, and a longer pattern taking precedence over a shorter pattern. If the same identifier value is requested to both be excluded and included the behavior is undefined; the service MAY return or omit the specified vocabulary but MUST NOT raise an exception.

Example 3: a Prefer header requesting all annotations within a metadata document to be returned

Prefer: include-annotations="*"

Example 4: a Prefer header requesting that no annotations are returned

Prefer: include-annotations="-*"

Example 5: a Prefer header requesting that all annotations defined under the "display" namespace (recursively) be returned

Prefer: include-annotations="display.*"

Example 6: a Prefer header requesting that the annotation with the term name subject within the display namespace be returned

Prefer: include-annotations="display.subject"

Example 7: a Prefer header requesting that all annotations defined under the "display" namespace (recursively) with the qualifier “tablet” be returned

Prefer: include-annotations="display.*#tablet"

The include-annotations preference is only a hint to the service. The service MAY ignore the preference and is free to decide whether or not to return annotations not specified in the include-annotations preference.

In the case that the client has specified the include-annotations preference in the request, the service SHOULD include a Preference-Applied response header containing the include-annotations preference to specify the annotations actually included, where applicable, in the response. This value may differ from the annotations requested in the Prefer header of the request.

If the include-annotations preference is specified on an individual request within a batch, then it specifies the preference for that individual request. Individual requests within a batch that don’t include the include-annotations preference inherit the preference of the overall batch request.

Note: The include-annotations preference was named odata.include-annotations in OData version 4.0. Services that support the include-annotations preference SHOULD also support odata.include-annotations for OData 4.0 clients and clients SHOULD use odata.include-annotations for compatibility with OData 4.0 services. If both include-annotations and odata.include-annotations preferences are specified in the same request, the value of the include-annotations preference SHOULD be used.

8.2.8.5 Preference maxpagesize (odata.maxpagesize)

The maxpagesize preference is used to request that each collection within the response contain no more than the number of items specified as the positive integer value of this preference. The syntax of the maxpagesize preference is defined in [OData-ABNF].

Example 8: a request for customers and their orders would result in a response containing one collection with customer entities and for every customer a separate collection with order entities. The client could specify maxpagesize=50 in order to request that each page of results contain a maximum of 50 customers, each with a maximum of 50 orders.

If a collection within the result contains more than the specified maxpagesize, the collection SHOULD be a partial set of the results with a next link to the next page of results. The client MAY specify a different value for this preference with every request following a next link.

In the example given above, the result page should include a next link for the customer collection, if there are more than 50 customers, and additional next links for all returned orders collections with more than 50 entities.

If the client has specified the maxpagesize preference in the request, and the service limits the number of items in collections within the response through server-driven paging, the service MAY include a Preference-Applied response header containing the maxpagesize preference and the maximum page size applied. This value may differ from the value requested by the client.

The maxpagesize preference SHOULD NOT be applied to a batch request, but MAY be applied to individual requests within a batch.

Note: The maxpagesize preference was named odata.maxpagesize in OData version 4.0. Services that support the maxpagesize preference SHOULD also support odata.maxpagesize for OData 4.0 clients and clients SHOULD use odata.maxpagesize for compatibility with OData 4.0 services. If both maxpagesize and odata.maxpagesize preferences are specified in the same request, the value of the maxpagesize preference SHOULD be used.

8.2.8.6 Preference omit-values

The omit-values preference specifies values that MAY be omitted from a response payload. Valid values are nulls or defaults.

If nulls is specified, then the service MAY omit properties containing null values from the response, in which case it MUST specify the Preference-Applied response header with omit-values=nulls.

If defaults is specified, then the service MAY omit properties containing default values from the response, including nulls for properties that have no other defined default value. Nulls MUST be included for properties that have a non-null default value defined. If the service omits default values, it MUST specify the Preference-Applied response header with omit-values=defaults.

Properties with instance annotations are not affected by this preference and MUST be included in the payload if they would be included without this preference. Clients MUST NOT try to reconstruct a null or default value for properties for which an instance annotation is present and no property value is present, for example if the property is omitted due to permissions and has been replaced with the instance annotation Core.Permissions and a value of None, see [OData-VocCore].

Properties with null or default values MUST be included in delta payloads, if modified.

The response to a POST operation MUST include any properties not set to their default value, and the response to a PUT/PATCH operation MUST include any properties whose values were changed as part of the operation.

The omit-values preference does not affect a request payload.

8.2.8.7 Preference return=representation and return=minimal

The return=representation and return=minimal preferences are defined in [RFC7240].

In OData, return=representation or return=minimal is defined for use with a POST, PUT, or PATCH Data Modification Request, or with an Action Request. Specifying a preference of return=representation or return=minimal in a GET or DELETE request does not have any effect.

A preference of return=representation or return=minimal is allowed on an individual Data Modification Request or Action Request within a batch, subject to the same restrictions, but SHOULD return a 4xx Client Error if specified on the batch request itself.

A preference of return=minimal requests that the service invoke the request but does not return content in the response. The service MAY apply this preference by returning 204 No Content in which case it MAY include a Preference-Applied response header containing the return=minimal preference.

A preference of return=representation requests that the service invokes the request and returns the modified resource. The service MAY apply this preference by returning the representation of the successfully modified resource in the body of the response, formatted according to the rules specified for the requested format. In this case the service MAY include a Preference-Applied response header containing the return=representation preference.

The return preference SHOULD NOT be applied to a batch request, but MAY be applied to individual requests within a batch.

8.2.8.8 Preference respond-async

The respond-async preference, as defined in [RFC7240], allows clients to request that the service process the request asynchronously.

If the client has specified respond-async in the request, the service MAY process the request asynchronously and return a 202 Accepted response.

The respond-async preference MAY be used for batch requests, in which case it applies to the batch request as a whole and not to individual requests within the batch request.

In the case that the service applies the respond-async preference it MUST include a Preference-Applied response header containing the respond-async preference.

A service MAY specify the support for the respond-async preference using an annotation with term Capabilities.AsynchronousRequestsSupported, see [OData-VocCap].

Example 9: a service receiving the following header might choose to respond

· asynchronously if the synchronous processing of the request will take longer than 10 seconds

· synchronously after 5 seconds

· asynchronously (ignoring the wait preference)

· synchronously after 15 seconds (ignoring respond-async preference and the wait preference)

Prefer: respond-async, wait=10

8.2.8.9 Preference track-changes (odata.track-changes)

The track-changes preference is used to request that the service return a delta link that can subsequently be used to obtain changes (deltas) to this result. The syntax of the track-changes preference is defined in [OData-ABNF].

For paged results, the preference MUST be specified on the initial request. Services MUST ignore the track-changes preference if applied to the next link.

The delta link MUST NOT be returned prior to the final page of results.

The service includes a Preference-Applied response header in the first page of the response containing the track-changes preference to signal that changes are being tracked.

A service MAY specify the support for the track-changes preference using an annotation with term Capabilities.ChangeTracking, see [OData-VocCap].

The track-changes preference SHOULD NOT be applied to a batch request, but MAY be applied to individual requests within a batch.

Note: The track-changes preference was named odata.track-changes in OData version 4.0. Services that support the track-changes preference SHOULD also support odata.track-changes for OData 4.0 clients and clients SHOULD use odata.track-changes for compatibility with OData 4.0 services.

8.2.8.10 Preference wait

The wait preference, as defined in [RFC7240], is used to establish an upper bound on the length of time, in seconds, the client is prepared to wait for the service to process the request synchronously once it has been received.

If the respond-async preference is also specified, the client requests that the service respond asynchronously after the specified length of time.

If the respond-async preference has not been specified, the service MAY interpret the wait as a request to timeout after the specified period of time.

If the wait preference is specified on an individual request within a batch, then it specifies the maximum amount of time to wait for that individual request. If the wait preference is specified on a batch, then it specifies the maximum time to wait for the entire batch.

8.3 Response Headers

In addition to the Common Headers, the following response headers have defined meaning in OData.

8.3.1 Header AsyncResult

A 4.01 service MUST include the AsyncResult header in 200 OK response from a status monitor resource in order to indicate the final HTTP Response Status Code of an asynchronously executed request.

The AsyncResult header SHOULD NOT be applied to individual responses within a batch.

8.3.2 Header ETag

A response MAY include an ETag header, see [RFC7232]. Services MUST include this header if they require an ETag to be specified when modifying the resource.

Services MUST support specifying the value returned in the ETag header in an If-None-Match header of a subsequent Data Request for the resource. Clients MUST specify the value returned in the ETag header, or star (*), in an If-Match header of a subsequent Data Modification Request or Action Request in order to apply optimistic concurrency control in updating, deleting, or invoking an action bound to the resource.

As OData allows multiple formats for representing the same structured information, services SHOULD use weak ETags that only depend on the representation-independent entity state. A strong ETag MUST change whenever the representation of an entity changes, so it has to depend on the Content-Type, the Content-Encoding, and potentially other characteristics of the response.

An ETag header MAY also be returned on a metadata document request or service document request to allow the client subsequently to make a conditional request for the metadata or service document. Clients can also compare the value of the ETag header returned from a metadata document request to the metadata ETag returned in a response in order to verify the version of the metadata used to generate that response.

The ETag header SHOULD NOT be included for the overall batch response, but MAY be included in individual responses within a batch.

8.3.3 Header Location

The Location header MUST be returned in the response from a Create Entity or Create Media Entity request to specify the edit URL, or for read-only entities the read URL, of the created entity, and in responses returning 202 Accepted to specify the URL that the client can use to request the status of an asynchronous request.

The Location header SHOULD NOT be included for the overall batch response, but MAY be included in individual responses within a batch.

8.3.4 Header OData-EntityId

A response to a create or upsert operation that returns 204 No Content MUST include an OData-EntityId response header. The value of the header is the entity-id of the entity that was acted on by the request. The syntax of the OData-EntityId header is defined in [OData-ABNF].

The OData-EntityID header SHOULD NOT be included for the overall batch response, but MAY be included in individual responses within a batch.

8.3.5 Header OData-Error

A response with an in-stream error MAY include an OData-Error trailing header if the transport protocol supports trailing headers (e.g. HTTP/1.1 with chunked transfer encoding, or HTTP/2).

The value of this trailing header is a standard OData error response according to the OData response format, encoded suitably for transport in a header, see e.g. [OData-JSON].

8.3.6 Header Preference-Applied

In a response to a request that specifies a Prefer header, a service MAY include a Preference-Applied header, as defined in [RFC7240], specifying how individual preferences within the request were handled.

The value of the Preference-Applied header is a comma-separated list of preferences applied in the response. For more information on the individual preferences, see the Prefer header.

If the Preference-Applied header is specified on an individual response within a batch, then it specifies the preferences applied to that individual response. If the Preference-Applied header is specified on a batch response, then it specifies the preferences applied to the overall batch.

8.3.7 Header Retry-After

A service MAY include a Retry-After header, as defined in [RFC7231], in 202 Accepted and in 3xx Redirect responses

The Retry-After header specifies the duration of time, in seconds, that the client is asked to wait before retrying the request or issuing a request to the resource returned as the value of the Location header.

8.3.8 Header Vary

If a response varies depending on the OData-Version of the response, the service MUST include a Vary header listing the OData-MaxVersion request header field to allow correct caching of the response.

If a response varies depending on the applied preferences (allow-entityreferences, include-annotations, omit-values, return), the service MUST include a Vary header listing the Prefer request header field to allow correct caching of the response.

Alternatively, the server MAY include a Vary header with the special value * as defined by [RFC7231], Section 8.2.1. Note that this will make it impossible for a proxy to cache the response, see [RFC7240].

9 Common Response Status Codes

An OData service MAY respond to any request using any valid HTTP status code appropriate for the request. A service SHOULD be as specific as possible in its choice of HTTP status codes.

The following represent the most common success response codes. In some cases, a service MAY respond with a more specific success code.

9.1 Success Responses

The following response codes represent successful requests.

9.1.1 Response Code 200 OK

A request that does not create a resource returns 200 OK if it is completed successfully and the value of the resource is not null. In this case, the response body MUST contain the value of the resource specified in the request URL.

9.1.2 Response Code 201 Created

A Create Entity, Create Media Entity, or Invoke Action request that successfully creates a resource returns 201 Created. In this case, the response body MUST contain the resource created.

9.1.3 Response Code 202 Accepted

202 Accepted indicates that the Data Service Request has been accepted and has not yet completed executing asynchronously. The asynchronous handling of requests is defined in the sections on Asynchronous Requests and Asynchronous Batch Requests..

9.1.4 Response Code 204 No Content

A request returns 204 No Content if the requested resource has the null value, or if the service applies a return=minimal preference. In this case, the response body MUST be empty.

As defined in [RFC7231], a Data Modification Request that responds with 204 No Content MAY include an ETag header with a value reflecting the result of the data modification if and only if the client can reasonably “know” the new representation of the resource without actually receiving it. For a PUT request this means that the response body of a corresponding 200 OK or 201 Created response would have been identical to the request body, i.e. no server-side modification of values sent in the request body, no server-calculated values etc. For a PATCH request this means that the response body of a corresponding 200 OK or 201 Created response would have consisted of all values sent in the request body, plus (for values not sent in the request body) server-side values corresponding to the ETag value sent in the If-Match header of the PATCH request, i.e. the previous values “known” to the client.

9.1.5 Response Code 3xx Redirection

As per [RFC7231], a 3xx Redirection indicates that further action needs to be taken by the client in order to fulfill the request. In this case, the response SHOULD include a Location header, as appropriate, with the URL from which the result can be obtained; it MAY include a Retry-After header.

9.1.6 Response Code 304 Not Modified

As per [RFC7232], a 304 Not Modified is returned when the client performs a GET request containing an If-None-Match header and the content has not changed. In this case the response SHOULD NOT include other headers in order to prevent inconsistencies between cached entity-bodies and updated headers.

The service MUST ensure that no observable change has occurred to the state of the service as a result of any request that returns a 304 Not Modified.

9.2 Client Error Responses

Error codes in the 4xx range indicate a client error, such as a malformed request.

The service MUST ensure that no observable change has occurred to the state of the service as a result of any request that returns an error status code.

In the case that a response body is defined for the error code, the body of the error is as defined for the appropriate format.

9.2.1 Response Code 404 Not Found

404 Not Found indicates that the resource specified by the request URL does not exist. The response body MAY provide additional information.

9.2.2 Response Code 405 Method Not Allowed

405 Method Not Allowed indicates that the resource specified by the request URL does not support the request method. In this case the response MUST include an Allow header containing a list of valid request methods for the requested resource as defined in [RFC7231].

9.2.3 Response Code 406 Not Acceptable

406 Not Acceptable indicates that the resource specified by the request URL does not have a current representation that would be acceptable for the client according to the request headers Accept, Accept-Charset, and Accept-Language, and that the service is unwilling to supply a default representation.

9.2.4 Response Code 410 Gone

410 Gone indicates that the requested resource is no longer available. This can happen if a client has waited too long to follow a delta link or a status-monitor-resource link, or a next link on a collection that was requested with snapshot isolation.

9.2.5 Response Code 412 Precondition Failed

As defined in [RFC7232], 412 Precondition Failed indicates that the client has performed a conditional request and the resource fails the condition. The service MUST ensure that no observable change occurs as a result of the request.

9.2.6 Response Code 424 Failed Dependency

424 Failed Dependency indicates that a request was not performed due to a failed dependency; for example, a request within a batch that depended upon a request that failed.

9.3 Server Error Responses

As defined in [RFC7231], error codes in the 5xx range indicate service errors.

9.3.1 Response Code 501 Not Implemented

If the client requests functionality not implemented by the OData Service, the service responds with 501 Not Implemented and SHOULD include a response body describing the functionality not implemented.

9.4 Error Response Body

The representation of an error response body is format-specific. It consists at least of the following information:

· code: required non-null, non-empty, language-independent string. Its value is a service-defined error code. This code serves as a sub-status for the HTTP error code specified in the response.

· message: required non-null, non-empty, language-dependent, human-readable string describing the error. The Content-Language header MUST contain the language code from [RFC5646] corresponding to the language in which the value for message is written.

· target: optional nullable, potentially empty string indicating the target of the error, for example, the name of the property in error.

· details: optional, potentially empty collection of structured instances with code, message, and target following the rules above.

· innererror: optional structured instance with service-defined content.

Service implementations SHOULD carefully consider which information to include in production environments to guard against potential security concerns around information disclosure.

9.5 In-Stream Errors

In the case that the service encounters an error after sending a success status to the client, the service MUST leave the response malformed according to its content-type. Clients MUST treat the entire response as being in error.

Services MAY include the header OData-Error as a trailing header if supported by the transport protocol (e.g. HTTP/1.1 with chunked transfer encoding, or HTTP/2).

10 Context URL

The context URL describes the content of the payload. It consists of the canonical metadata document URL and a fragment identifying the relevant portion of the metadata document. The context URL makes response payloads “self-contained”, allowing a recipient to retrieve metadata, resolve references, and construct canonical links omitted from response payloads in certain optimized formats.

Request payloads generally do not require context URLs as the type of the payload can generally be determined from the request URL.

For details on how the context URL is used to describe a payload, see the relevant sections in the particular format.

The following subsections describe how the context URL is constructed for each category of payload by providing a context URL template. The context URL template uses the following terms:

· {context-url} is the canonical resource path to the $metadata document,

· {entity-set} is the name of an entity set or path to a containment navigation property,

· {entity} is the canonical URL for an entity,

· {singleton} is the canonical URL for a singleton entity,

· {select-list} is an optional parenthesized comma-separated list of selected properties, instance annotations, functions, and actions,

· {property-path} is the path to a structural property of the entity,

· {type-name} is a qualified type name,

· {/type-name} is an optional type-cast segment containing the qualified name of a derived or implemented type prefixed with a forward slash.

The full grammar for the context URL is defined in [OData-ABNF]. Note that the syntax of the context URL is independent of whatever URL conventions the service uses for addressing individual entities.

10.1 Service Document

Context URL template:

{context-url}

The context URL of the service document is the metadata document URL of the service.

Example 10: resource URL and corresponding context URL

http://host/service/

http://host/service/$metadata

10.2 Collection of Entities

Context URL template:

{context-url}#{entity-set}

{context-url}#Collection({type-name})

If all entities in the collection are members of one entity set, its name is the context URL fragment.

Example 11: resource URL and corresponding context URL

http://host/service/Customers

http://host/service/$metadata#Customers

If the entities are contained, then entity-set is the top-level entity set or singleton followed by the path to the containment navigation property of the containing entity.

Example 12: resource URL and corresponding context URL for contained entities

http://host/service/Orders(4711)/Items

http://host/service/$metadata#Orders(4711)/Items

If the entities in the response are not bound to a single entity set, such as from a function or action with no entity set path, a function import or action import with no specified entity set, or a navigation property with no navigation property binding, the context URL specifies the type of the returned entity collection.

10.3 Entity

Context URL template:

{context-url}#{entity-set}/$entity

{context-url}#{type-name}

If a response or response part is a single entity of the declared type of an entity set, /$entity is appended to the context URL.

Example 13: resource URL and corresponding context URL

http://host/service/Customers(1)

http://host/service/$metadata#Customers/$entity

If the entity is contained, then entity-set is the canonical URL for the containment navigation property of the containing entity, e.g. Orders(4711)/Items.

Example 14: resource URL and corresponding context URL for contained entity

http://host/service/Orders(4711)/Items(1)

http://host/service/$metadata#Orders(4711)/Items/$entity

If the response is not bound to a single entity set, such as an entity returned from a function or action with no entity set path, a function import or action import with no specified entity set, or a navigation property with no navigation property binding, the context URL specifies the type of the returned entity.

10.4 Singleton

Context URL template:

{context-url}#{singleton}

If a response or response part is a singleton, its name is the context URL fragment.

Example 15: resource URL and corresponding context URL

http://host/service/MainSupplier

http://host/service/$metadata#MainSupplier

10.5 Collection of Derived Entities

Context URL template:

{context-url}#{entity-set}{/type-name}

If an entity set consists exclusively of derived entities, a type-cast segment is added to the context URL.

Example 16: resource URL and corresponding context URL

http://host/service/Customers/Model.VipCustomer

http://host/service/$metadata#Customers/Model.VipCustomer

10.6 Derived Entity

Context URL template:

{context-url}#{entity-set}{/type-name}/$entity

If a response or response part is a single entity of a type derived from the declared type of an entity set, a type-cast segment is appended to the entity set name.

Example 17: resource URL and corresponding context URL

http://host/service/Customers(2)/Model.VipCustomer

http://host/service/$metadata#Customers/Model.VipCustomer/$entity

10.7 Collection of Projected Entities

Context URL templates:

{context-url}#{entity-set}{/type-name}{select-list}

{context-url}#Collection({type-name}){select-list}

If a result contains only a subset of properties, the parenthesized comma-separated list of the selected defined or dynamic properties, instance annotations, navigation properties, functions, and actions is appended to the context URL representing the collection of entities.

Regardless of how contained structural properties are represented in the request URL (as paths or as select options) they are represented in the context URL using path syntax, as defined in OData 4.0. Similarly, expands nested within $select are represented the same as expands outside of $select.

The shortcut * represents the list of all structural properties. Properties defined on types derived from the declared type of the entity set (or type specified in the type-cast segment if specified) are prefixed with the qualified name of the derived type as defined in [OData-ABNF].

The list also contains explicitly selected or expanded instance annotations. It is possible to select or expand only instance annotations, in which case only those selected or expanded annotations appear in the result. Note that annotations specified only in the include-annotations preference do not appear in the context URL and do not affect the selected/expanded properties.

Operations in the context URL are represented using the namespace- or alias-qualified name. Function names suffixed with parentheses represent a specific overload, while function names without parentheses represent all overloads of the function.

OData 4.01 responses MAY use the shortcut pattern {namespace}.* to represent the list of all bound actions or functions available for entities in the collection, see system query option $select.

Example 18: resource URL and corresponding context URL

http://host/service/Customers?$select=Address,Orders

http://host/service/$metadata#Customers(Address,Orders)

10.8 Projected Entity

Context URL templates:

{context-url}#{entity-set}{/type-name}{select-list}/$entity

{context-url}#{singleton}{select-list}

{context-url}#{type-name}{select-list}

If a single entity contains a subset of properties, the parenthesized comma-separated list of the selected defined or dynamic properties, instance annotations, navigation properties, functions, and actions is appended to the {entity-set} after an optional type-cast segment and prior to appending /$entity. If the response is not a subset of a single entity set, the {select-list} is instead appended to the {type-name} of the returned entity.

The shortcut * represents the list of all structural properties. Properties defined on types derived from the type of the entity set (or type specified in the type-cast segment if specified) are prefixed with the qualified name of the derived type as defined in [OData-ABNF]. Note that expanded properties are automatically included in the response.

OData 4.01 responses MAY use the shortcut pattern {namespace}.* to represent the list of all bound actions or functions available for the returned entity, see system query option $select.

Example 19: resource URL and corresponding context URL

http://host/service/Customers(1)?$select=Name,Rating

http://host/service/$metadata#Customers(Name,Rating)/$entity

10.9 Collection of Expanded Entities

Context URL template:

{context-url}#{entity-set}{/type-name}{select-list}

{context-url}#Collection({type-name}){select-list}

For a 4.01 response, if a navigation property is explicitly expanded, then in addition to any non-suffixed names of any selected properties, navigation properties, functions or actions, the comma-separated list of properties MUST include the name of the expanded property, suffixed with the parenthesized comma-separated list of any properties of the expanded navigation property that are selected or expanded. If the expanded navigation property does not contain a nested $select or $expand, then the expanded property is suffixed with empty parentheses. If the expansion is recursive for nested children, a plus sign (+) is infixed between the navigation property name and the opening parenthesis.

For a 4.0 response, the expanded navigation property suffixed with parentheses MAY be omitted from the select-list if it does not contain a nested $select or $expand, but MUST still be present, without a suffix, if it is explicitly selected.

If the context URL includes only expanded navigation properties (i.e., only navigation properties suffixed with parentheses), then all structural properties are implicitly selected (same as if there were no properties listed in the select-list).

Navigation properties with expanded references are not represented in the context URL.

Example 20: resource URL and corresponding context URL - select and expand

http://host/service/Customers?$select=Name&$expand=Address/Country

http://host/service/$metadata#Customers(Name,Address/Country())

Example 21: resource URL and corresponding context URL – expand $ref

http://host/service/Customers?$expand=Orders/$ref

http://host/service/$metadata#Customers

Example 22: resource URL and corresponding context URL – expand with $levels

http://host/service/Employees/Sales.Manager?$select=DirectReports
&$expand=DirectReports($select=FirstName,LastName;$levels=4)

http://host/service/$metadata
#Employees/Sales.Manager(DirectReports,
DirectReports+(FirstName,LastName))

10.10 Expanded Entity

Context URL template:

{context-url}#{entity-set}{/type-name}{select-list}/$entity

{context-url}#{singleton}{select-list}

{context-url}#{type-name}{select-list}

For a 4.01 response, if a navigation property is explicitly expanded, then in addition to the non-suffixed names of any selected properties, navigation properties, functions or actions, the comma-separated list of properties MUST include the name of the expanded property, suffixed with the parenthesized comma-separated list of any properties of the expanded navigation property that are selected or expanded. If the expanded navigation property does not contain a nested $select or $expand, then the expanded property is suffixed with empty parentheses. If the expansion is recursive for nested children, a plus sign (+) is infixed between the navigation property name and the opening parenthesis.

Navigation properties with expanded references are not represented in the context URL.

Example 23: resource URL and corresponding context URL

http://host/service/Employees(1)/Sales.Manager?
$expand=DirectReports($select=FirstName,LastName;$levels=4)

http://host/service/$metadata
#Employees/Sales.Manager(DirectReports+(FirstName,LastName))/$entity

10.11 Collection of Entity References

Context URL template:

{context-url}#Collection($ref)

If a response is a collection of entity references, the context URL does not contain the type of the referenced entities.

Example 24: resource URL and corresponding context URL for a collection of entity references

http://host/service/Customers('ALFKI')/Orders/$ref

http://host/service/$metadata#Collection($ref)

10.12 Entity Reference

Context URL template:

{context-url}#$ref

If a response is a single entity reference, $ref is the context URL fragment.

Example 25: resource URL and corresponding context URL for a single entity reference

http://host/service/Orders(10643)/Customer/$ref

http://host/service/$metadata#$ref

10.13 Property Value

Context URL templates:

{context-url}#{entity}/{property-path}{select-list}

{context-url}#{type-name}{select-list}

If a response represents an individual property of an entity with a canonical URL, the context URL specifies the canonical URL of the entity and the path to the structural property of that entity. The path MUST include cast segments for properties defined on types derived from the expected type of the previous segment.

If the property value does not contain explicitly or implicitly selected navigation properties or operations, OData 4.01 responses MAY use the less specific second template.

Example 26: resource URL and corresponding context URL

http://host/service/Customers(1)/Addresses

http://host/service/$metadata#Customers(1)/Addresses

10.14 Collection of Complex or Primitive Types

Context URL template:

{context-url}#Collection({type-name}){select-list}

If a response is a collection of complex types or primitive types that do not represent an individual property of an entity with a canonical URL, the context URL specifies the fully qualified type of the collection.

Example 27: resource URL and corresponding context URL

http://host/service/TopFiveHobbies()

http://host/service/$metadata#Collection(Edm.String)

10.15 Complex or Primitive Type

Context URL template:

{context-url}#{type-name}{select-list}

If a response is a complex type or primitive type that does not represent an individual property of an entity with a canonical URL, the context URL specifies the fully qualified type of the result.

Example 28: resource URL and corresponding context URL

http://host/service/MostPopularName()

http://host/service/$metadata#Edm.String

10.16 Operation Result

Context URL templates:

{context-url}#{entity-set}{/type-name}{select-list}

{context-url}#{entity-set}{/type-name}{select-list}/$entity

{context-url}#{entity}/{property-path}{select-list}

{context-url}#Collection({type-name}){select-list}

{context-url}#{type-name}{select-list}

If the response from an action or function is a collection of entities or a single entity that is a member of an entity set, the context URL identifies the entity set. If the response from an action or function is a property of a single entity, the context URL identifies the entity and property. Otherwise, the context URL identifies the type returned by the operation. The context URL will correspond to one of the former examples.

Example 29: resource URL and corresponding context URL

http://host/service/TopFiveCustomers()

http://host/service/$metadata#Customers

10.17 Delta Payload Response

Context URL template:

{context-url}#{entity-set}{/type-name}{select-list}/$delta

{context-url}#{entity}{select-list}/$delta

{context-url}#{entity}/{property-path}{select-list}/$delta

#$delta

The context URL of a delta response is the context URL of the response to the defining query, followed by /$delta. This includes singletons, single-valued navigation properties, and collection-valued navigation properties.

If the entities are contained, then {entity-set} is the top-level entity set followed by the path to the containment navigation property of the containing entity.

Example 30: resource URL and corresponding context URL

http://host/service/Customers?$deltatoken=1234

http://host/service/$metadata#Customers/$delta

The context URL of an update request body for a collection of entities is simply the fragment #$delta.

10.18 Item in a Delta Payload Response

Context URL templates:

{context-url}#{entity-set}/$deletedEntity

{context-url}#{entity-set}/$link

{context-url}#{entity-set}/$deletedLink

In addition to new or changed entities which have the canonical context URL for an entity, a delta response can contain deleted entities, new links, and deleted links. They are identified by the corresponding context URL fragment. {entity-set} corresponds to the set of the deleted entity, or source entity for an added or deleted link.

10.19 $all Response

Context URL template:

{context-url}#Collection(Edm.EntityType)

Responses to requests to the virtual collection $all (see [OData‑URL]) use the built-in abstract entity type. Each single entity in such a response has its individual context URL that identifies the entity set or singleton.

10.20 $crossjoin Response

Context URL template:

{context-url}#Collection(Edm.ComplexType)

Responses to requests to the virtual collections $crossjoin(...) (see [OData‑URL]) use the built-in abstract complex type. Single instances in these responses do not have a context URL.

11 Data Service Requests

This chapter describes the semantics of the HTTP verbs GET, POST, PATCH, PUT, and DELETE for OData resources.

GET requests:

· 11.1 Metadata Requests and subsections

· 11.2 Requesting Data and subsections

· 11.3 Requesting Changes and subsections

· 11.5.4 Functions and subsections

POST requests:

· 11.4.2 Create an Entity and subsections

· 11.4.7.1 Create a Media Entity

· 11.4.11 Positional Inserts

· 11.5.5 Actions and subsections

· 11.7 Batch Requests and subsections

PATCH and PUT requests:

· 11.4.3 Update an Entity and subsections

· 11.4.4 Upsert an Entity

· 11.4.6 Modifying Relationships between Entities and subsections

· 11.4.7.2 Update a Media Entity Stream

· 11.4.8.1 Update Stream Values

· 11.4.9.1 Update a Primitive Property

· 11.4.9.3 Update a Complex Property

· 11.4.9.4 Update a Collection Property

· 11.4.10 Managing Members of an Ordered Collection

· 11.4.12 Update a Collection of Entities

· 11.4.13 Update Members of a Collection

DELETE requests:

· 11.4.5 Delete an Entity

· 11.4.7.3 Delete a Media Entity

· 0 Delete Stream Values

· 11.4.9.2 Set a Value to Null

· 11.4.14 Delete Members of a Collection

11.1 Metadata Requests

An OData service is a self-describing service that exposes metadata defining the entity sets, singletons, relationships, entity types, and operations.

11.1.1 Service Document Request

Service documents enable simple hypermedia-driven clients to enumerate and explore the resources offered by the data service.

OData services MUST support returning a service document from the root URL of the service (the service root).

The format of the service document is dependent upon the format selected.

11.1.2 Metadata Document Request

An OData metadata document is a representation of the data model that describes the data and operations exposed by an OData service.

[OData-CSDLJSON] describes a JSON representation for OData metadata documents and provides a JSON schema to validate their contents. The media type of the JSON representation of an OData metadata document is application/json.

[OData-CSDLXML] describes an XML representation for OData metadata documents and provides an XML schema to validate their contents. The media type of the XML representation of an OData metadata document is application/xml.

OData services can expose a metadata document that describes the data model exposed by the service. The metadata document URL MUST be the root URL of the service with $metadata appended. To retrieve this document the client issues a GET request to the metadata document URL.

If a request for metadata does not specify a format preference (via Accept header or $format) then the XML representation MUST be returned.

11.2 Requesting Data

OData services support requests for data via HTTP GET requests.

The path of the URL specifies the target of the request (for example; the collection of entities, entity, navigation property, structural property, or operation). Additional query operators, such as filter, sort, page, and projection operations are specified through query options.

This section describes the types of data requests defined by OData. For complete details on the syntax for building requests, see [OData‑URL].

OData services are hypermedia driven services that return URLs to the client. If a client subsequently requests the advertised resource and the URL has expired, then the service SHOULD respond with 410 Gone. If this is not feasible, the service MUST respond with 404 Not Found.

The format of the returned data is dependent upon the request and the format specified by the client, either in the Accept header or using the $format query option. If the client specifies neither an Accept header nor the $format query option, the service is allowed to return the response in any format.

11.2.1 System Query Options

OData defines a number of system query options that allow refining the request. System query options are prefixed with the dollar ($) character, which is optional in OData 4.01. 4.01 services MUST support case-insensitive system query option names specified with or without the $ prefix.
Clients that want to work with 4.0 services MUST use lower case names and specify the $ prefix.

The result of the request MUST be as if the system query options were evaluated in the following order.

· $schemaversion MUST be evaluated first, because it may influence any further processing.

Prior to applying any server-driven paging:

$apply – defined in [OData-Aggregation]
$compute
$search
$filter
$count
$orderby
$skip
$top

After applying any server-driven paging:

$expand
$select
$format

11.2.2 Requesting Individual Entities

To retrieve an individual entity, the client makes a GET request to a URL that identifies the entity, e.g. its read URL.

The read URL can be obtained from a response payload containing that instance, for example as a readLink or editLink in an [OData-JSON] payload. In addition, services MAY support conventions for constructing a read URL using the entity's key value(s), as described in [OData‑URL].

The set of structural or navigation properties to return may be specified through $select or $expand system query options.

Clients MUST be prepared to receive additional properties in an entity or complex type instance that are not advertised in metadata, even for types not marked as open.

Properties that are not available, for example due to permissions, are not returned. In this case, the Core.Permissions annotation, defined in [OData-VocCore] MUST be returned for the property with a value of None.

If no entity exists with the specified request URL, the service responds with 404 Not Found.

11.2.3 Requesting the Media Stream of a Media Entity using $value

A media entity is an entity that represents an out-of-band stream, such as a photograph.

Use a media entity if the out-of-band stream is the main topic of interest and the media entity is just structured additional information attached to the stream. Use a normal entity with one or more stream properties if the structured data of the entity is the main topic of interest and the stream data is just additional information attached to the structured data.

To address the media stream represented by a media entity, clients append /$value to the resource path of the media entity URL. Services may redirect from this canonical URL to the source URL of the media stream.

Appending /$value to an entity that is not a media entity returns 400 Bad Request.

Attempting to retrieve the media stream from a single-valued navigation property referencing a media entity whose value is null returns 404 Not Found.

11.2.4 Requesting Individual Properties

To retrieve an individual property, the client issues a GET request to the property URL. The property URL is the entity read URL with "/" and the property name appended.

For complex typed properties, the path can be further extended with the name of an individual property of the complex type.

See [OData‑URL] for details.

If the property is single-valued and has the null value, the service responds with 204 No Content.

If the property is not available, for example due to permissions, the service responds with 404 Not Found.

Example 31:

GET http://host/service/Products(1)/Name

11.2.4.1 Requesting a Property's Raw Value using $value

To retrieve the raw value of a primitive type property, the client sends a GET request to the property value URL. See the [OData‑URL] document for details.

The Content-Type of the response is determined using the Accept header and the $format system query option.

The default format for single primitive values except Edm.Binary and the Edm.Geo types is text/plain.

The default format for Edm.Geo types is text/plain using the WKT (well-known text) format, see rules fullCollectionLiteral, fullLineStringLiteral, fullMultiPointLiteral, fullMultiLineStringLiteral, fullMultiPolygonLiteral, fullPointLiteral, and fullPolygonLiteral in [OData-ABNF].

The default format for Edm.Binary is the format specified by the Core.MediaType annotation of this property (see [OData-VocCore]) if this annotation is present. If not annotated, the format cannot be predicted by the client.

A $value request for a property that is null results in a 204 No Content response.

If the property is not available, for example due to permissions, the service responds with 404 Not Found.

Example 32:

GET http://host/service/Products(1)/Name/$value

11.2.5 Specifying Properties to Return

The $select and $expand system query options enable the client to specify the set of structural properties and navigation properties to include in a response. The service MAY include additional properties not specified in $select and $expand, including properties not defined in the metadata document.

11.2.5.1 System Query Option $select

The $select system query option requests that the service return only the properties, dynamic properties, actions and functions explicitly requested by the client. The service returns the specified content, if available, along with any available expanded navigation or stream properties, and MAY return additional information.

The value of the $select query option is a comma-separated list of properties, qualified action names, qualified function names, the star operator (*), or the star operator prefixed with the namespace or alias of the schema in order to specify all operations defined in the schema. Only aliases defined in the metadata document of the service can be used in URLs.

Example 33: request only the Rating and ReleaseDate for the matching Products

GET http://host/service/Products?$select=Rating,ReleaseDate

It is also possible to request all structural properties, including any dynamic properties, using the star operator. The star operator SHOULD NOT introduce navigation properties, actions or functions not otherwise requested.

Example 34:

GET http://host/service/Products?$select=*

Properties of related entities can be specified by including the $select query option within the $expand.

Example 35:

GET http://host/service/Products?$expand=Category($select=Name)

The properties specified in $select are represented in addition to any expanded navigation or stream properties. If a navigation property is specified in $select, then the corresponding navigation link is represented in the response. If the navigation property also appears in an $expand query option, then it is additionally represented as inline content.

Example 36: for each category, return the CategoryName and the Products navigation link

GET http://host/service/Categories?$select=CategoryName,Products

It is also possible to request all actions or functions available for each returned entity.

Example 37:

GET http://host/service/Products?$select=DemoService.*

Query options can be applied to a selected property by appending a semicolon-separated list of query options, enclosed in parentheses, to the property. Allowed system query options are $select and $compute for complex properties, plus $filter, $search, $count, $orderby, $skip, and $top for collection-valued properties. A property MUST NOT have select options specified in more than one place in a request and MUST NOT have both select options and expand options specified.

If the $select query option is not specified, the service returns the full set of properties or a default set of properties. The default set of properties MUST include all key properties. Services may change the default set of properties returned. This includes returning new properties by default and omitting properties previously returned by default. Clients that rely on specific properties in the response MUST use $select with the required properties or with *.

If the service returns less than the full set of properties, either because the client specified a select or because the service returned a subset of properties in the absence of a select, the context URL MUST reflect the set of selected properties and projected expanded navigation properties.

11.2.5.2 System Query Option $expand

The $expand system query option indicates the related entities and stream values that MUST be represented inline. The service MUST return the specified content, and MAY choose to return additional information.

The value of the $expand query option is a comma-separated list of navigation property names, stream property names, or $value indicating the stream content of a media-entity.

For navigation properties, the navigation property name is optionally followed by a /$ref path segment or a /$count path segment, and optionally a parenthesized set of expand options (for filtering, sorting, selecting, paging, or expanding the related entities).

For a full description of the syntax used when building requests, see [OData‑URL].

Example 38: for each customer entity within the Customers entity set the value of all related Orders will be represented inline

GET http://host/service.svc/Customers?$expand=Orders

Example 39: for each customer entity within the Customers entity set the references to the related Orders will be represented inline

GET http://host/service.svc/Customers?$expand=Orders/$ref

Example 40: for each customer entity within the Customers entity set the media stream representing the customer photo will be represented inline

GET http://host/service.svc/Customers?$expand=Photo

11.2.5.2.1 Expand Options

The set of expanded entities can be further refined through the application of expand options, expressed as a semicolon-separated list of system query options, enclosed in parentheses, see [OData‑URL].

Allowed system query options are $filter, $select, $orderby, $skip, $top, $count, $search, $expand, $compute, and $levels.

Example 41: for each customer entity within the Customers entity set, the value of those related Orders whose Amount is greater than 100 will be represented inline

GET http://host/service.svc/Customers?$expand=Orders($filter=Amount gt 100)

Example 42: for each order within the Orders entity set, the following will be represented inline:

· The Items related to the Orders identified by the resource path section of the URL and the products related to each order item.

The Customer related to each order returned.

GET http://host/service.svc/Orders?$expand=Items($expand=Product),Customer

Example 43: for each customer entity in the Customers entity set, the value of all related InHouseStaff will be represented inline if the entity is of type VipCustomer or a subtype of that. For entities that are not of type VipCustomer, or any of its subtypes, that entity may be returned with no inline representation for the expanded navigation property InHouseStaff (the service can always send more than requested)

GET http://host/service.svc/Customers?$expand=SampleModel.VipCustomer/InHouseStaff

11.2.5.2.1.1 Expand Option $levels

The $levels expand option can be used to specify the number of levels of recursion for a hierarchy in which the related entity type is the same as, or can be cast to, the source entity type. A $levels option with a value of 1 specifies a single expand with no recursion. The same expand options are applied at each level of the hierarchy.

Services MAY support the symbolic value max in addition to numeric values. In that case they MUST solve circular dependencies by injecting an entity reference somewhere in the circular dependency.

Clients using $levels=max MUST be prepared to handle entity references in cases were a circular reference would occur otherwise.

4.01 services that support max SHOULD do so in a case-insensitive manner. Clients that want to work with 4.0 services MUST use lower case.

Example 44: return each employee from the Employees entity set and, for each employee that is a manager, return all direct reports, recursively to four levels

GET http://host/service/Employees?$expand=Model.Manager/DirectReports($levels=4)

11.2.5.3 System Query Option $compute

The $compute system query option allows clients to define computed properties that can be used in a $select or within a $filter or $orderby expression.

Computed properties SHOULD be included as dynamic properties in the result and MUST be included if $select is specified with the computed property name, or star (*).

Example 45: compute total price for order items (line breaks only for readability)

GET http://host/service/Customers?
$filter=Orders/any(o:o/TotalPrice gt 100)
&$expand=Orders($compute=Price mult Qty as TotalPrice

;$select=Name,Price,Qty,TotalPrice)

11.2.6 Querying Collections

OData services support querying collections of entities, complex type instances, and primitive values.

The target collection is specified through a URL, and query operations such as filter, sort, paging, and projection are specified as system query options optionally prefixed with a dollar ($) character. 4.01 Services MUST support case-insensitive system query option names specified with or without the $ prefix. Clients that want to work with 4.0 services MUST use lower case names and specify the $ prefix.

The same system query option MUST NOT be specified more than once for any resource.

An OData service MAY support some or all of the system query options defined. If a data service does not support a system query option, it MUST fail any request that contains the unsupported option and SHOULD return 501 Not Implemented.

11.2.6.1 System Query Option $filter

The $filter system query option restricts the set of items returned.

Example 46: return all Products whose Price is less than $10.00

GET http://host/service/Products?$filter=Price lt 10.00

The $count segment may be used within a $filter expression to limit the items returned based on the exact count of related entities or items within a collection-valued property.

Example 47: return all Categories with less than 10 products

GET http://host/service/Categories?$filter=Products/$count lt 10

The value of the $filter option is a Boolean expression as defined in [OData-ABNF].

11.2.6.1.1 Built-in Filter Operations

OData supports a set of built-in filter operations, as described in this section.

4.01 services MUST support case-insensitive operation names. Clients that want to work with 4.0 services MUST use lower case operation names.

For a full description of the syntax used when building requests, see [OData‑URL].

Operator	Description	Example
Comparison Operators
eq	Equal	Address/City eq 'Redmond'
ne	Not equal	Address/City ne 'London'
gt	Greater than	Price gt 20
ge	Greater than or equal	Price ge 10
lt	Less than	Price lt 20
le	Less than or equal	Price le 100
has	Has flags	Style has Sales.Color'Yellow'
in	Is a member of	Address/City in ('Redmond', 'London')
Logical Operators
and	Logical and	Price le 200 and Price gt 3.5
or	Logical or	Price le 3.5 or Price gt 200
not	Logical negation	not endswith(Description,'milk')
Arithmetic Operators
add	Addition	Price add 5 gt 10
sub	Subtraction	Price sub 5 gt 10
mul	Multiplication	Price mul 2 gt 2000
div	Division	Price div 2 gt 4
divby	Decimal Division	Price divby 2 gt 3.5
mod	Modulo	Price mod 2 eq 0
Grouping Operators
( )	Precedence grouping	(Price sub 5) gt 10

11.2.6.1.2 Built-in Query Functions

OData supports a set of built-in functions that can be used within $filter operations. The following table lists the available functions.

4.01 services MUST support case-insensitive built-in function names. Clients that want to work with 4.0 services MUST use lower case names.

For a full description of the syntax used when building requests, see [OData‑URL].

OData does not define an ISNULL or COALESCE operator. Instead, there is a null literal that can be used in comparisons.

Function	Example
String and Collection Functions
concat	concat(concat(City,', '), Country) eq 'Berlin, Germany'
contains	contains(CompanyName,'freds')
endswith	endswith(CompanyName,'Futterkiste')
indexof	indexof(CompanyName,'lfreds') eq 1
length	length(CompanyName) eq 19
startswith	startswith(CompanyName,’Alfr’)
substring	substring(CompanyName,1) eq 'lfreds Futterkiste'
Collection Functions
hassubset	hassubset([4,1,3],[3,1])
hassubsequence	hassubsequence([4,1,3,1],[1,1])
String Functions
matchesPattern	matchesPattern(CompanyName,'%5EA.*e$')
tolower	tolower(CompanyName) eq 'alfreds futterkiste'
toupper	toupper(CompanyName) eq 'ALFREDS FUTTERKISTE'
trim	trim(CompanyName) eq 'Alfreds Futterkiste'
Date and Time Functions
day	day(StartTime) eq 8
date	date(StartTime) ne date(EndTime)
fractionalseconds	second(StartTime) eq 0
hour	hour(StartTime) eq 1
maxdatetime	EndTime eq maxdatetime()
mindatetime	StartTime eq mindatetime()
minute	minute(StartTime) eq 0
month	month(BirthDate) eq 12
now	StartTime ge now()
second	second(StartTime) eq 0
time	time(StartTime) le StartOfDay
totaloffsetminutes	totaloffsetminutes(StartTime) eq 60
totalseconds	totalseconds(duration'PT1M') eq 60
year	year(BirthDate) eq 0
Arithmetic Functions
ceiling	ceiling(Freight) eq 33
floor	floor(Freight) eq 32
round	round(Freight) eq 32
Type Functions
cast	cast(ShipCountry,Edm.String)
isof	isof(NorthwindModel.Order)
isof	isof(ShipCountry,Edm.String)
Geo Functions
geo.distance	geo.distance(CurrentPosition,TargetPosition)
geo.intersects	geo.intersects(Position,TargetArea)
geo.length	geo.length(DirectRoute)
Conditional Functions
case	case(X gt 0:1,X lt 0:-1,true:0)

11.2.6.1.3 Parameter Aliases

Parameter aliases can be used in place of literal values in entity keys, function parameters, or within a $compute, $filter or $orderby expression. Parameters aliases are names beginning with an at sign (@).

Actual parameter values are specified as query options in the query part of the request URL. The query option name is the name of the parameter alias, and the query option value is the value to be used for the specified parameter alias.

Example 48: returns all employees whose Region property matches the string parameter value "WA"

GET http://host/service.svc/Employees?$filter=Region eq @p1&@p1='WA'

Parameter aliases allow the same value to be used multiple times in a request and may be used to reference primitive, structured, or collection values.

If a parameter alias is not given a value in the query part of the request URL, the value MUST be assumed to be null. A parameter alias can be used in multiple places within a request URL, but its value MUST NOT be specified more than once.

Parameter alias values used in /$filter path segments are always passed as expressions (because that is the expected type of the parameter).

All other parameter alias values are evaluated in the context of the resource identified by the path segment in which they are assigned and passed as values into the expression. Parameter alias value assignments MAY be nested within $expand and $select, in which case they are evaluated relative to the resource context of the $expand or $select.

Example 49: returns all employees, expands their manager, and expands all direct reports with the same first name as the manager, using a parameter alias for $this to pass the manager into the filter on the expanded direct reports

GET http://host/service.svc/Employees?$expand=Manager(@m=$this;$expand=DirectReports($filter=@m/FirstName eq FirstName))

11.2.6.2 System Query Option $orderby

The $orderby System Query option specifies the order in which items are returned from the service.

The value of the $orderby System Query option contains a comma-separated list of expressions whose primitive result values are used to sort the items. A special case of such an expression is a property path terminating on a primitive property. A type cast using the qualified entity type name is required to order by a property defined on a derived type. Only aliases defined in the metadata document of the service can be used in URLs.

The expression can include the suffix asc for ascending or desc for descending, separated from the property name by one or more spaces. If asc or desc is not specified, the service MUST order by the specified property in ascending order. 4.01 services MUST support case-insensitive values for asc and desc. Clients that want to work with 4.0 services MUST use lower case values.

Null values come before non-null values when sorting in ascending order and after non-null values when sorting in descending order.

Items are sorted by the result values of the first expression, and then items with the same value for the first expression are sorted by the result value of the second expression, and so on.

The Boolean value false comes before the value true in ascending order.

Services SHOULD order language-dependent strings according to the content-language of the response, and SHOULD annotate string properties with language-dependent order with the term Core.IsLanguageDependent, see [OData-VocCore].

Values of type Edm.Stream or any of the Geo types cannot be sorted.

Example 50: return all Products ordered by release date in ascending order, then by rating in descending order

GET http://host/service/Products?$orderby=ReleaseDate asc, Rating desc

Related entities may be ordered by specifying $orderby within the $expand clause.

Example 51: return all Categories, and their Products ordered according to release date and in descending order of rating

GET http://host/service/Categories?
$expand=Products($orderby=ReleaseDate asc, Rating desc)

$count may be used within a $orderby expression to order the returned items according to the exact count of related entities or items within a collection-valued property.

Example 52: return all Categories ordered by the number of Products within each category

GET http://host/service/Categories?$orderby=Products/$count

11.2.6.3 System Query Option $top

The $top system query option specifies a non-negative integer n that limits the number of items returned from a collection. The service returns the number of available items up to but not greater than the specified value n.

Example 53: return only the first five products of the Products entity set

GET http://host/service/Products?$top=5

If no unique ordering is imposed through an $orderby query option, the service MUST impose a stable ordering across requests that include $top.

11.2.6.4 System Query Option $skip

The $skip system query option specifies a non-negative integer n that excludes the first n items of the queried collection from the result. The service returns items starting at position n+1.

Example 54: return products starting with the 6th product of the Products entity set

GET http://host/service/Products?$skip=5

Where $top and $skip are used together, $skip MUST be applied before $top, regardless of the order in which they appear in the request.

Example 55: return the third through seventh products of the Products entity set

GET http://host/service/Products?$top=5&$skip=2

If no unique ordering is imposed through an $orderby query option, the service MUST impose a stable ordering across requests that include $skip.

11.2.6.5 System Query Option $count

The $count system query option with a value of true specifies that the total count of items within a collection matching the request be returned along with the result.

Example 56: return, along with the results, the total number of products in the collection

GET http://host/service/Products?$count=true

The count of related entities can be requested by specifying the $count query option within the $expand clause.

Example 57:

GET http://host/service/Categories?$expand=Products($count=true)

A $count query option with a value of false (or not specified) hints that the service SHOULD NOT return a count.

The service returns an HTTP Status code of 400 Bad Request if a value other than true or false is specified.

The $count system query option ignores any $top, $skip, or $expand query options, and returns the total count of results across all pages including only those results matching any specified $filter and $search. Clients should be aware that the count returned inline may not exactly equal the actual number of items returned, due to latency between calculating the count and enumerating the last value or due to inexact calculations on the service.

How the count is encoded in the response body is dependent upon the selected format.

11.2.6.6 System Query Option $search

The $search system query option restricts the result to include only those items matching the specified search expression. The definition of what it means to match is dependent upon the implementation.

Example 58: return all Products that match the search term "bike"

GET http://host/service/Products?$search=bike

The search expression can contain phrases, enclosed in double-quotes.

Example 59: return all Products that match the phrase "mountain bike"

GET http://host/service/Products?$search="mountain bike"

The upper-case keyword NOT restricts the set of entities to those that do not match the specified term.

Example 60: return all Products that do not match "clothing"

GET http://host/service/Products?$search=NOT clothing

Multiple terms within a search expression are separated by a space (implicit AND) or the upper-case keyword AND, indicating that all such terms must be matched.

Example 61: return all Products that match both "mountain" and "bike"

GET http://host/service/Products?$search=mountain AND bike

The upper-case keyword OR is used to return entities that satisfy either the immediately preceding or subsequent expression.

Example 62: return all Products that match either "mountain" or "bike"

GET http://host/service/Products?$search=mountain OR bike

Parentheses within the search expression group together multiple expressions.

Example 63: return all Products that match either "mountain" or "bike" and do not match clothing

GET http://host/service/Products?$search=(mountain OR bike) AND NOT clothing

The operations within a search expression MUST be evaluated in the following order: grouping operator, NOT operator, AND operator, OR operator

If both $search and $filter are specified in the same request, only those items satisfying both criteria are returned.

The value of the $search option is a search expression as defined in [OData-ABNF].

11.2.6.7 Server-Driven Paging

Responses that include only a partial set of the items identified by the request URL MUST contain a link that allows retrieving the next partial set of items. This link is called a next link; its representation is format-specific. The final partial set of items MUST NOT contain a next link.

The client can request a maximum page size through the maxpagesize preference. The service may apply this requested page size or implement a page size different than, or in the absence of, this preference.

OData clients MUST treat the URL of the next link as opaque, and MUST NOT append system query options to the URL of a next link. Services may not allow a change of format on requests for subsequent pages using the next link. Clients therefore SHOULD request the same format on subsequent page requests using a compatible Accept header. OData services may use the reserved system query option $skiptoken when building next links. Its content is opaque, service-specific, and must only follow the rules for URL query parts.

OData clients MUST NOT use the system query option $skiptoken when constructing requests.

11.2.6.8 Requesting an Individual Member of an Ordered Collection

Individual members of collections of primitive and complex types annotated with the Ordered term (see [OData-VocCore]) are addressable by appending a segment containing the zero-based ordinal to the URL of the collection. A negative ordinal indexes from the end of the collection, with -1 representing the last item in the collection.

Entities are stably addressable using their canonical URL and are not accessible using an ordinal index.

Example 64: the first address in a list of addresses for MainSupplier

GET http://host/service/Suppliers(MainSupplier)/Addresses/0

11.2.7 Requesting Related Entities

To request related entities according to a particular relationship, the client issues a GET request to the source entity’s request URL, followed by a forward slash and the name of the navigation property representing the relationship.

If the navigation property does not exist on the entity indicated by the request URL, the service returns 404 Not Found.

If the relationship terminates on a collection, the response MUST be the format-specific representation of the collection of related entities. If no entities are related, the response is the format-specific representation of an empty collection.

If the relationship terminates on a single entity, the response MUST be the format-specific representation of the related single entity. If no entity is related, the service returns 204 No Content.

Example 65: return the supplier of the product with ID=1 in the Products entity set

GET http://host/service/Products(1)/Supplier

11.2.8 Requesting Entity References

To request entity references in place of the actual entities, the client issues a GET request with /$ref appended to the resource path.

If the resource path does not identify an entity or a collection of entities, the service returns 404 Not Found.

If the resource path identifies a collection, the response MUST be the format-specific representation of a collection of entity references pointing to the related entities. If no entities are related, the response is the format-specific representation of an empty collection. The response MAY contain an ETag header for the collection whose value changes if the collection of references changes, i.e. a reference is added or removed.

If the resource path identifies a single existing entity, the response MUST be the format-specific representation of an entity reference. The response MAY contain an ETag header which represents the identity of the referenced entity. If the resource path terminates in a single-valued navigation path, the ETag value changes if the relationship is changed and points to a different OData entity. If the resource path is the canonical path for a single entity, the returned ETag can never change.

If the resource path terminates on a single entity and no such entity exists, the service returns either 204 No Content or 404 Not Found.

Example 66: collection with an entity reference for each Order related to the Product with ID=0

GET http://host/service/Products(0)/Orders/$ref

11.2.9 Resolving an Entity-Id

To resolve an entity-id, e.g. obtained in an entity reference, into a representation of the identified entity, the client issues a GET request to the $entity resource located at the URL $entity relative to the service root. The entity-id MUST be specified using the system query option $id.

Example 67: return the entity representation for a given entity-id

GET http://host/service/$entity?$id=http://host/service/Products(0)

A type segment following the $entity resource casts the resource to the specified type. If the identified entity is not of the specified type, or a type derived from the specified type, the service returns 404 Not Found.

After applying a type-cast segment to cast to a specific type, the system query options $select and $expand can be specified in GET requests to the $entity resource.

Example 68: return the entity representation for a given entity-id and specify properties to return

GET http://host/service/$entity/Model.Customer
?$id=http://host/service/Customers('ALFKI')

&$select=CompanyName,ContactName

&$expand=Orders

11.2.10 Requesting the Number of Items in a Collection

To request only the number of items of a collection of entities or items of a collection-valued property, the client issues a GET request with /$count appended to the resource path of the collection.

On success, the response body MUST contain the exact count of items matching the request after applying any $filter or $search system query options, formatted as a simple primitive integer value with media type text/plain. Clients SHOULD NOT combine the system query options $top, $skip, $orderby, $expand, and $format with the path suffix /$count. The result of such a request is undefined.

Example 69: return the number of products in the Products entity set

GET http://host/service/Products/$count

With 4.01 services the /$count segment MAY be used in combination with the /$filter path segment to count the items in the filtered collection.

Example 70: return the number of products whose Price is less than $10.00

GET http://host/service/Products/$filter(@foo)/$count?@foo=Price lt 10.00

For backwards compatibility, the /$count suffix MAY be used in combination with the $filter system query option.

Example 71: return the number of products whose Price is less than $10.00

GET http://host/service/Products/$count?$filter=Price lt 10.00

The $filter system query option MUST NOT be used in conjunction with a both a /$count path segment and a /$filter path segment.

The /$count suffix can also be used in path expressions within system query options, e.g. $filter.

Example 72: return all customers with more than five interests

GET http://host/service/Customers?$filter=Interests/$count gt 5

Example 73: return all categories with more than one product over $5.00

GET http://host/service/Categories?
$filter=Products/$filter(Price gt 5.0)/$count gt 1

11.2.11 System Query Option $format

The $format system query option specifies the media type of the response.

The $format query option, if present in a request, MUST take precedence over the value(s) specified in the Accept request header.

The value of the $format system query option is a valid internet media type, optionally including parameters.

In addition, format-specific abbreviations may be used, e.g. json for application/json, see [OData-JSON], but format parameters MUST NOT be appended to the format abbreviations.

Example 74: the request

GET http://host/service/Orders?$format=application/json;metadata=full

is equivalent to a request with an Accept header using the same media type; it requests the set of Order entities represented using the JSON media type including full metadata, as defined in [OData-JSON].

Example 75: the request

GET http://host/service/Orders?$format=json

is equivalent to a request with the Accept header set to application/json; it requests the set of Order entities represented using the JSON media type with minimal metadata, as defined in [OData-JSON].

In metadata document requests, the values application/xml and application/json, along with their subtypes and parameterized variants, as well as the format-specific abbreviations xml and json, are reserved for this specification.

11.2.12 System Query Option $schemaversion

The $schemaversion system query option MAY be included in any request. For a metadata document request the value of the $schemaversion system query option addresses a specific schema version. For all other request types the value specifies the version of the schema against which the request is made. The syntax of the $schemaversion system query option is defined in [OData-ABNF].

The value of the $schemaversion system query option MUST be a version of the schema as returned in the Core.SchemaVersion annotation, defined in [OData-VocCore], of a previous request to the metadata document, or * in order to specify the current version of the metadata.

If specified, the service MUST process the request according to the specified version of the metadata.

Clients can retrieve the current version of the metadata by making a metadata document request with a $schemaversion system query option value of *, and SHOULD include the value from the returned Core.SchemaVersion annotation in the $schemaversion system query option of subsequent requests.

If the $schemaversion system query option is not specified in a request for the metadata document, the service MUST return a version of the metadata with no breaking changes over time, and the processing of all other requests that omit the $schemaversion system query option MUST be compatible with that “unversioned” schema. For more information on breaking changes, see Model Versioning.

If the $schemaversion system query option is specified on an individual request within a batch, then it specifies the version of the schema to apply to that individual request. Individual requests within a batch that don’t include the $schemaversion system query option inherit the schema version of the overall batch request.

If the $schemaversion system query option is specified, but the version of the schema doesn’t exist, the request is answered with a response code 404 Not Found. The response body SHOULD provide additional information.

11.3 Requesting Changes

Services advertise their change-tracking capabilities by annotating entity sets with the Capabilities.ChangeTracking term defined in [OData-VocCap].

Any GET request to retrieve one or more entities MAY allow change-tracking.

Clients request that the service track changes to a result by specifying the track-changes preference on a request. If supported for the request, the service includes a Preference-Applied header in the response containing the track-changes preference and includes a delta link in a result for a single entity, and on the last page of results for a collection of entities.

11.3.1 Delta Links

Delta links are opaque, service-generated links that the client uses to retrieve subsequent changes to a result.

Delta links are based on a defining query that describes the set of results for which changes are being tracked; for example, the request that generated the results containing the delta link. The delta link encodes the collection of entities for which changes are being tracked, along with a starting point from which to track changes. OData services may use the reserved system query option $deltatoken when building delta links. Its content is opaque, service-specific, and must only follow the rules for URL query parts.

If the defining query contains a $schemaversion system query option, the response MUST be represented according to that schema version.

If the defining query contains a $filter or $search, the response MUST include only changes to entities matching the specified criteria. Added entities MUST be returned for entities that were added or changed and now match the specified criteria, and deleted entities MUST be returned for entities that are changed to no longer match the criteria of $filter or $search.

The delta link MUST NOT encode any client top or skip value, and SHOULD NOT encode a request for an inline count.

If the defining query includes expanded relationships, the delta link MUST return changes, additions, or deletions to the expanded entities, as well as added or deleted links to expanded entities or nested collections representing current membership. This is independent of whether related entity themselves or only references to related entities were requested.

Navigation properties specified in the $select list of a defining query are not used to define the scope or contents of the items being tracked. Clients can specify /$ref in $expand in order to specify interest in the set of related entities without interest in changes to the content of those related entities.

If an expanded entity becomes orphaned because all paths to the entity as specified in the defining query have been broken (i.e. due to relationship changes and/or changes or deletions to parent entities) then the service MUST return the appropriate notifications for the client to determine that the entity has been orphaned (i.e. the changed relationships and removed parent entities). The client should not assume that it will receive additional notifications for such an orphaned entity.

Entities are considered changed if any of the structural properties have changed. Changes to related entities and to streams are not considered a change to the entity containing the stream or navigation property.

If the defining query contains a projection, the generated delta link SHOULD logically include the same projection, such that the delta query only includes fields specified in the projection. Services MAY use the projection to limit the entities returned to those that have changed within the selected fields, but the client MUST be prepared to receive entities returned whether or not the field that changed was specified in the projection.

11.3.2 Using Delta Links

The client requests changes by invoking the GET method on the delta link. The client MUST NOT attempt to append system query options to the delta link. The Accept header MAY be used to specify the desired response format.

Clients SHOULD specify the same Accept-Language header when querying the delta link as was specified in the defining query. Services MAY return 406 Not Acceptable if a different Accept-Language is specified. If a service does support an Accept-Language header it MAY return changes only visible in that language, or MAY include records that have changes not visible in the requested language.

The /$count segment can be appended to the path of a delta link in order to get just the number of changes available. The count includes all added, changed, or deleted entities, as well as added or deleted links.

The results of a request against the delta link may span multiple pages but MUST be ordered by the service across all pages in such a way as to guarantee consistency when applied in order to the response which contained the delta link.

Services SHOULD return only changed entities, but MAY return additional entities matching the defining query for which the client may not see a change.

In order to continue tracking changes beyond the current set, the client specifies track-changes on the initial request to the delta link but is not required to repeat it for subsequent pages. The new delta link appears at the end of the last page of changes and MUST return all changes subsequent to the last change of the previous delta link.

If no changes have occurred, the response is an empty collection that contains a delta link for subsequent changes if requested. This delta link MAY be identical to the delta link resulting in the empty collection of changes.

If the delta link is no longer valid, the service responds with 410 Gone, and SHOULD include the URL for refetching the entire set in the Location header of the response.

11.3.3 Delta Payloads

A delta payload represents changes to a known state. A delta payload includes added entities, changed entities, and deleted entities, as well as a representation of added and removed relationships.

Delta payloads can be requested from the service using a delta link or provided as updates to the service.

11.4 Data Modification

Updatable OData services support Create, Update, and Delete operations for some or all exposed entities. Additionally, Actions supported by a service can affect the state of the system.

A successfully completed Data Modification Request must not violate the integrity of the data.

The client may request whether content be returned from a Create, Update, or Delete request, or the invocation of an Action, by specifying the return Prefer header.

11.4.1 Common Data Modification Semantics

Data Modification Requests share the following semantics.

11.4.1.1 Use of ETags for Avoiding Update Conflicts

Each entity has its own ETag value that MUST change when structural properties or links from that entity have changed. In addition, modifying, adding, or deleting a contained entity MAY change the ETag of the parent entity.

Collections of entities (including collections of related entities) MAY have their own ETag value whose semantics is service-specific. It typically changes if entities are added to or removed from the collection, or if an entity in the collection is changed. The ETag of a collection of related entities reached via a navigation property MAY differ from the ETag of the entity containing the navigation property.

A Data Modification Request on an existing resource or an Action Request invoking an action bound to an existing resource MAY require optimistic concurrency control. Services SHOULD announce this via annotations with the terms Core.OptimisticConcurrency in [OData-VocCore] and Capabilities.NavigationRestrictions (nested property OptimisticConcurrencyControl) in [OData-VocCap].

If optimistic concurrency control is required for a resource, the service MUST include an ETag header in a response to a GET request to the resource, and MAY include the ETag in a format-specific manner in responses containing that resource.

The presence of an ETag header in a response does not imply in itself that the resource requires optimistic concurrency control; the ETag may just be used for caching and/or conditional GET requests.

If an ETag value is specified in an If-Match or If-None-Match header of a Data Modification Request or Action Request, the operation MUST only be invoked if the If-Match or If-None-Match condition is satisfied.

If the client does not specify an If-Match request header in a Data Modification Request or Action Request on a resource that requires optimistic concurrency control, the service responds with a 428 Precondition Required and MUST ensure that no observable change occurs as a result of the request. Clients can attempt to disable optimistic concurrency control by specifying If-Match with a value of *. Services MAY reject such requests.

For requests including an OData-Version header value of 4.01, any ETag values specified in the request body of an update request MUST be * or match the current value for the record being updated.

11.4.1.2 Handling of DateTimeOffset Values

Services SHOULD preserve the offset of Edm.DateTimeOffset values, if possible. However, where the underlying storage does not support offset services may be forced to normalize the value to some common time zone (i.e. UTC) in which case the result would be returned with that time zone offset. If the service normalizes values, it MUST fail evaluation of the query functions year, month, day, hour, and time for literal values that are not stated in the time zone of the normalized values.

11.4.1.3 Handling of Properties Not Advertised in Metadata

Clients MUST be prepared to receive additional properties in an entity or complex type instance that are not advertised in metadata, even for types not marked as open. By using PATCH when updating entities, clients can ensure that such properties values are not lost if omitted from the update request.

11.4.1.4 Handling of Integrity Constraints

Services may impose cross-entity integrity constraints. Certain referential constraints, such as requiring an entity to be created with related entities can be satisfied through creating or linking related entities when creating the entity. Other constraints might require multiple changes to be processed in an all-or-nothing fashion.

11.4.1.5 Returning Results from Data Modification Requests

Clients can request whether created or modified resources are returned from create, update, and upsert operations using the return preference header. In the absence of such a header, services SHOULD return the created or modified content unless the resource is a stream property value.

When returning content other than for an update to a media entity stream, services MUST return the same content as a subsequent request to retrieve the same resource. For updating media entity streams, the content of a non-empty response body MUST be the updated media entity.

Requests that return a single instance of a structured type or a collection of structured type instances MAY specify the system query options $expand and $select.

Requests that return a collection MAY specify the system query option $filter.

If one or more of these query options are present, this implies a return=representation preference if no return preference is specified.

If one or more of these query options are present with a return=minimal preference, the service SHOULD NOT return a representation and MUST include a Preference-Applied header if it does not return a representation.

If one or more of these query options are present and the service returns a representation, then the service MUST apply the specified query options. If it cannot apply the specified query options appropriately, it MUST NOT fail the request solely due to the presence of these query options and instead MUST return 204 No Content.

11.4.2 Create an Entity

To create an entity in a collection, the client sends a POST request to that collection's URL. The POST body MUST contain a single valid entity representation.

The entity representation MAY include references to existing entities as well as content for new related entities, but MUST NOT contain content for existing related entities. The result of the operation is the entity with relationships to all included references to existing entities as well as all related entities created inline. If the key properties for an entity include key properties of a directly related entity, those related entities MUST be included either as references to existing entities or as content for new related entities.

An entity may also be created as the result of an Upsert operation.

If the target URL for the collection is a navigation link, the new entity is automatically linked to the entity containing the navigation link.

If the target URL terminates in a type cast segment, then the segment MUST specify the type of, or a type derived from, the type of the collection, and the entity MUST be created as that specified type.

To create an open entity (an instance of an open type), additional property values beyond those specified in the metadata MAY be sent in the request body. The service MUST treat these as dynamic properties and add them to the created instance.

If the entity being created is not an open entity, additional property values beyond those specified in the metadata SHOULD NOT be sent in the request body. The service MUST fail if unable to persist all property values specified in the request.

Properties computed by the service (annotated with the term Core.Computed, see [OData-VocCore]) and properties that are tied to properties of the principal entity by a referential constraint, can be omitted and MUST be ignored if included in the request.

Properties with a defined default value, nullable properties, and collection-valued properties omitted from the request are set to the default value, null, or an empty collection, respectively.

Upon successful completion, the response MUST contain a Location header that contains the edit URL or read URL of the created entity.

Upon successful completion the service MUST respond with either 201 Created and a representation of the created entity, or 204 No Content if the request included a Prefer header with a value of return=minimal and did not include the system query options $select and $expand.

11.4.2.1 Link to Related Entities When Creating an Entity

To create a new entity with links to existing entities in a single request, the client includes references to the related entities in the request body.

The representation for referencing related entities is format-specific.

Example 76: using the JSON format, 4.0 clients can create a new manager entity with links to two existing employees by applying the odata.bind annotation to the DirectReports navigation property

{

"@odata.type":"#Northwind.Manager",

"ID": 1,

"FirstName": "Pat",

"LastName": "Griswold",

"DirectReports@odata.bind": [

"http://host/service/Employees(5)",

"http://host/service/Employees(6)"

]

}

Example 77: using the JSON format, 4.01 clients can create a new manager entity with links to two existing employees by including the entity-ids within the DirectReports navigation property

{

"@type":"#Northwind.Manager",

"ID": 1,

"FirstName": "Pat",

"LastName": "Griswold",

"DirectReports": [

{"@id": "Employees(5)"},

{"@id": "Employees(6)"}

]

}

Upon successful completion of the operation, the service creates the requested entity and relates it to the requested existing entities.

If the target URL for the collection the entity is created in and binding information provided in the POST body contradicts the implicit binding information provided by the request URL, the request MUST fail, and the service responds with 400 Bad Request.

Upon failure of the operation, the service MUST NOT create the new entity. In particular, the service MUST never create an entity in a partially valid state (with the navigation property unset).

11.4.2.2 Create Related Entities When Creating an Entity

A request to create an entity that includes related entities, represented using the appropriate inline representation, is referred to as a “deep insert”.

Media entities MUST contain the base64url-encoded representation of their media stream as a virtual property $value when nested within a deep insert.

Each included related entity is processed observing the rules for creating an entity as if it was posted against the original target URL extended with the navigation path to this related entity.

On success, the service MUST create all entities and relate them. If the service responds with 201 Created, the response MUST be expanded to at least the level that was present in the deep-insert request.

Clients MAY associate an id with individual nested entities in the request by using the Core.ContentID term defined in [OData-VocCore]. Services that respond with 201 Created SHOULD annotate the entities in the response using the same Core.ContentID value as specified in the request. Services SHOULD advertise support for deep inserts, including support for returning the Core.ContentID, through the Capabilities.DeepInsertSupport term, defined in [OData-VocCap]; services that advertise support through Capabilities.DeepInsertSupport MUST return the Core.ContentID for the inserted or updated entities.

The continue-on-error preference is not supported for deep insert operations.

On failure, the service MUST NOT create any of the entities.

11.4.3 Update an Entity

To update an individual entity, the client makes a PATCH or PUT request to a URL that identifies the entity. Services MAY restrict updates only to requests addressing the edit URL of the entity.

Services SHOULD support PATCH as the preferred means of updating an entity. PATCH provides more resiliency between clients and services by directly modifying only those values specified by the client.

The semantics of PATCH, as defined in [RFC5789], is to merge the content in the request payload with the [entity’s] current state, applying the update only to those components specified in the request body. Collection properties and primitive properties provided in the payload corresponding to updatable properties MUST replace the value of the corresponding property in the entity or complex type. Missing properties of the containing entity or complex property, including dynamic properties, MUST NOT be directly altered unless as a side effect of changes resulting from the provided properties.

Services MAY additionally support PUT but should be aware of the potential for data-loss in round-tripping properties that the client may not know about in advance, such as open or added properties, or properties not specified in metadata. Services that support PUT MUST replace all values of structural properties with those specified in the request body. Missing non-key, updatable structural properties not defined as dependent properties within a referential constraint MUST be set to their default values. Omitting a non-nullable property with no service-generated or default value from a PUT request results in a 400 Bad Request error. Missing dynamic structural properties MUST be removed or set to null.

For requests with an OData-Version header with a value of 4.01 or greater, the media stream of a media entity can be updated by specifying the base64url-encoded representation of the media stream as a virtual property $value.

Updating a dependent property that is tied to a key property of the principal entity through a referential constraint updates the relationship to point to the entity with the specified key value. If there is no such entity, the update fails.

Updating a principle property that is tied to a dependent entity through a referential constraint on the dependent entity updates the dependent property.

Key and other properties marked as read-only in metadata (including computed properties), as well as dependent properties that are not tied to key properties of the principal entity, can be omitted from the request. If the request contains a value for one of these properties, the service MUST ignore that value when applying the update. Services MUST return an error if an insert or update contains a new value for a property marked as updatable that cannot currently be changed by the user (i.e., given the state of the object or permissions of the user). The service MAY return success in this case if the specified value matches the value of the property. Clients SHOULD use PATCH and specify only those properties intended to be changed.

Entity id and entity type cannot be changed when updating an entity. However, format-specific rules might in some cases require providing entity id and entity type values in the payload when applying the update.

For requests with an OData-Version header with a value of 4.01 or greater, if the entity representation in the request body includes an ETag value, the update MUST NOT be performed and SHOULD return 412 Precondition Failed if the supplied ETag value is not * and does not match the current ETag value for the entity. ETag values in request bodies MUST be ignored for requests containing an OData-Version header with a value of 4.0.

If an update specifies both a binding to a single-valued navigation property and a dependent property that is tied to a key property of the principal entity according to the same navigation property, then the dependent property is ignored, and the relationship is updated according to the value specified in the binding.

If the entity being updated is open, then additional values for properties beyond those specified in the metadata or returned in a previous request MAY be sent in the request body. The service MUST treat these as dynamic properties.

If the entity being updated is not open, then additional values for properties beyond those specified in the metadata or returned in a previous request SHOULD NOT be sent in the request body. The service MUST fail if it is unable to persist all updatable property values specified in the request.

Upon successful completion the service responds with either 200 OK and a representation of the updated entity, or 204 No Content. The client may request that the response SHOULD include a body by specifying a Prefer header with a value of return=representation, or by specifying the system query options $select or $expand. If the service uses ETags for optimistic concurrency control, the entities in the response MUST include ETags.

11.4.3.1 Update Related Entities When Updating an Entity

Update requests with an OData-Version header with a value of 4.0 MUST NOT contain related entities as inline content. Such requests MAY contain binding information for navigation properties. For single-valued navigation properties this replaces the relationship. For collection-valued navigation properties this adds to the relationship.

Payloads with an OData-Version header with a value of 4.01 or greater MAY include nested entities and entity references that specify the full set of to be related entities, or a nested delta payload representing the related entities that have been added, removed, or changed. Such a request is referred to as a “deep update”. If the nested collection is represented identical to an expanded navigation property, then the set of nested entities and entity references specified in a successful update request represents the full set of entities to be related according to that relationship and MUST NOT include added links, deleted links, or deleted entities.

Example 78: using the JSON format, a 4.01 PATCH request can update a manager entity. Following the update, the manager has three direct reports; two existing employees and one new employee named Suzanne Brown. The LastName of employee 6 is updated to Smith.

{

"@type":"#Northwind.Manager",

"FirstName" : "Patricia",

"DirectReports": [

{

"@id": "Employees(5}"

{

"@id": "Employees(6}",

"LastName": "Smith"

{

"FirstName": "Suzanne",

"LastName": "Brown"

}

]

}

If the nested collection is represented as a delta annotation on the navigation property, then the collection contains members to be added or changed and MAY include deleted entities for entities that are no longer part of the collection, using the delta payload format. If the deleted entity specifies a reason as deleted, then the entity is both removed from the collection and deleted, otherwise it is removed from the collection and only deleted if the relationship is contained. Non-key properties of the deleted entity are ignored. Nested collections MUST NOT contain added or deleted links. If the request contains nested delta collections, then the PATCH verb must be specified.

If a nested entity has the same id or key fields as an existing entity, the existing entity is updated according to the semantics of the PUT or PATCH request. Nested entities that have no id or key fields, or for which the id or key fields do not match existing entities, are treated as inserts. If the nested collection does not represent a containment relationship and has no navigation property binding, then such entities MUST include a context URL specifying the entity set in which the new entity is to be created. If any nested entities contain both id and key fields, they MUST identify the same entity, or the request is invalid.

Example 79: using the JSON format, a 4.01 PATCH request can specify a nested delta representation to:

· delete employee 3 and remove link to it

· remove the link to employee 4 and do not delete it

· add a link to employee 5

· change the last name of employee 6 and link to it if necessary

· add a new employee named “Suzanne Brown” and link to it

{

"@type": "#Northwind.Manager",

"FirstName": "Patricia",

"DirectReports@delta": [

{

"@removed": {

"reason": "deleted"

"@id": "Employees(3)"

{

"@removed": {

"reason": "changed"

"@id": "Employees(4)"

{

"@id": "Employees(5)"

{

"@id": "Employees(6)",

"LastName": "Smith"

{

"FirstName": "Suzanne",

"LastName": "Brown"

}

]

}

Clients MAY associate an id with individual nested entities in the request by using the Core.ContentID term defined in [OData-VocCore]. Services that respond with 200 OK SHOULD annotate the entities in the response using the same Core.ContentID value as specified in the request. Services SHOULD advertise support for deep updates, including support for returning the Core.ContentID, through the Capabilities.DeepUpdateSupport term, defined in [OData-VocCap].

The continue-on-error preference is not supported for deep update operations.

On failure, the service MUST NOT apply any of the changes specified in the request.

11.4.4 Upsert an Entity

An upsert occurs when the client sends an update request to a valid URL that identifies a single entity that does not yet exist. In this case the service MUST handle the request as a create entity request or fail the request altogether.

Upserts are not supported against media entities, single-valued non-containment navigation properties, or entities whose keys values are generated by the service. Services MUST fail an update request to a URL that would identify such an entity and the entity does not yet exist.

Singleton entities can be upserted if they are nullable. Services supporting this SHOULD advertise it by annotating the singleton with the term Capabilities.UpdateRestrictions (nested property Upsertable with value true) defined in [OData-VocCap].

Key and other non-updatable properties, as well as dependent properties that are not tied to key properties of the principal entity, MUST be ignored by the service in processing the Upsert request.

To ensure that an update request is not treated as an insert, the client MAY specify an If-Match header in the update request. The service MUST NOT treat an update request containing an If-Match header as an insert.

A PUT or PATCH request MUST NOT be treated as an update if an If-None-Match header is specified with a value of *.

11.4.5 Delete an Entity

To delete an individual entity, the client makes a DELETE request to a URL that identifies the entity. Services MAY restrict deletes only to requests addressing the edit URL of the entity.

The request body SHOULD be empty. Singleton entities can be deleted if they are nullable. Services supporting this SHOULD advertise it by annotating the singleton with the term Capabilities.DeleteRestrictions (nested property Deletable with value true) defined in [OData-VocCap].

On successful completion of the delete, the response MUST be 204 No Content and contain an empty body.

Services MUST implicitly remove relations to and from an entity when deleting it; clients need not delete the relations explicitly.

Services MAY implicitly delete or modify related entities if required by integrity constraints. If integrity constraints are declared in $metadata using a ReferentialConstraint element, services MUST modify affected related entities according to the declared integrity constraints, e.g. by deleting dependent entities, or setting dependent properties to null or their default value.

One such integrity constraint results from using a navigation property in a key definition of an entity type. If the related “key” entity is deleted, the dependent entity is also deleted.

11.4.6 Modifying Relationships between Entities

Relationships between entities are represented by navigation properties as described in Data Model. URL conventions for navigation properties are described in [OData‑URL].

11.4.6.1 Add a Reference to a Collection-Valued Navigation Property

A successful POST request to a navigation property's references collection adds a relationship to an existing entity. The request body MUST contain a single entity reference that identifies the entity to be added. See the appropriate format document for details.

On successful completion, the response MUST be 204 No Content and contain an empty body.

Note that if the two entities are already related prior to the request, the request is completed successfully.

11.4.6.2 Remove a Reference to an Entity

A successful DELETE request to the URL that represents a reference to a related entity removes the relationship to that entity.

In OData 4.0, the entity reference to be removed within a collection-valued navigation property is the URL that represents the collection of related references, with the reference to be removed identified by the $id query option. OData 4.01 services additionally support using the URL that represents the reference of the collection member to be removed, identified by key, as described in [OData‑URL].

For single-valued navigation properties, the $id query option MUST NOT be specified.

The DELETE request MUST NOT violate any integrity constraints in the data model.

On successful completion, the response MUST be 204 No Content and contain an empty body.

11.4.6.3 Change the Reference in a Single-Valued Navigation Property

A successful PUT request to a single-valued navigation property’s reference resource changes the related entity. The request body MUST contain a single entity reference that identifies the existing entity to be related. See the appropriate format document for details.

On successful completion, the response MUST be 204 No Content and contain an empty body.

Alternatively, a relationship MAY be updated as part of an update to the source entity by including the required binding information for the new target entity. This binding information is format-specific, see [OData-JSON] for details.

If the single-valued navigation property is used in the key definition of an entity type, it cannot be changed and the request MUST fail with 405 Method Not Allowed or an other appropriate error.

11.4.6.4 Replace all References in a Collection-valued Navigation Property

A successful PUT request to a collection-valued navigation property’s reference resource replaces the set of related entities. The request body MUST contain a collection of entity references in the same format as returned by a GET request to the navigation property’s reference resource.

A successful DELETE request to a collection-valued navigation property’s reference resource removes all related references from the collection.

11.4.7 Managing Media Entities

A media entity MUST have a source URL that can be used to read the media stream, and MAY have a media edit URL that can be used to write to the media stream.

Because a media entity has both a media stream and standard entity properties special handling is required.

11.4.7.1 Create a Media Entity

A POST request to a media entity's entity set creates a new media entity. The request body MUST contain the media value (for example, the photograph) whose media type MUST be specified in a Content-Type header. The request body is always interpreted as the media value, even if it has the media type of an OData format supported by the service. It is not possible to set the structural properties of the media entity when creating the media entity.

Upon successful completion, the response MUST contain Location header that contains the edit URL of the created media entity.

Upon successful completion the service responds with either 201 Created, or 204 No Content if the request included a Prefer header with a value of return=minimal.

11.4.7.2 Update a Media Entity Stream

A successful PUT request to the media edit URL of a media entity changes the media stream of the entity.

If the entity includes an ETag value for the media stream, the client MUST include an If-Match header with the ETag value.

The request body MUST contain the new media value for the entity whose media type MUST be specified in a Content-Type header.

On success, the service MUST respond with either 204 No Content and an empty body, or 200 OK if the client specified the preference return=representation, in which case the response body MUST contain the updated media entity.

11.4.7.3 Delete a Media Entity

A successful DELETE request to the entity's edit URL or to the edit URL of its media stream deletes the media entity as described in Delete an Entity.

Deleting a media entity also deletes the media associated with the entity.

11.4.8 Managing Stream Properties

An entity may have one or more stream properties. Stream properties are properties of type Edm.Stream.

The values for stream properties do not usually appear in the entity payload. Instead, the values are generally read or written through URLs.

11.4.8.1 Update Stream Values

A successful PUT request to the edit URL of a stream property changes the media stream associated with that property.

If the stream metadata includes an ETag value, the client SHOULD include an If-Match header with the ETag value.

The request body MUST contain the new media value for the stream whose media type MUST be specified in a Content-Type header. It may have a Content-Length of zero to set the stream data to empty.

Stream properties MAY specify a list of acceptable media types using an annotation with term Core.AcceptableMediaTypes, see [OData-VocCore].

Clients MAY change the association between a stream property and a media stream by modifying the edit URL or read URL of the stream property. Services supporting this SHOULD advertise it by annotating the stream property with the term Capabilities.MediaLocationUpdateSupported defined in [OData-VocCap].

11.4.8.2 Delete Stream Values

A successful DELETE request to the edit URL of a stream property attempts to set the property to null and results in an error if the property is non-nullable.

Attempting to request a stream property whose value is null results in 204 No Content.

11.4.9 Managing Values and Properties Directly

Values and properties can be explicitly addressed with URLs. The edit URL of a property is the edit URL of the entity appended with the path segment(s) specifying the individual property. The edit URL allows properties to be individually modified. See [OData‑URL] for details on addressing individual properties.

11.4.9.1 Update a Primitive Property

A successful PUT request to the edit URL for a primitive property updates the value of the property. The message body MUST contain the new value, formatted as a single property according to the specified format.

A successful PUT request to the edit URL for the raw value of a primitive property updates the property with the raw value specified in the payload. The payload MUST be formatted as an appropriate content type for the raw value of the property.

The same rules apply whether this is a regular property or a dynamic property.

Upon successful completion the service responds with either 200 OK or 204 No Content. The client may request that the response SHOULD include a body by specifying a Prefer header with a value of return=representation.

Services MUST return an error if the property is not updatable.

11.4.9.2 Set a Value to Null

A successful DELETE request to the edit URL for a structural property, or to the edit URL of the raw value of a primitive property, sets the property to null. The request body is ignored and should be empty.

A DELETE request to a non-nullable value MUST fail and the service respond with 400 Bad Request or other appropriate error.

The same rules apply whether the target is the value of a regular property or the value of a dynamic property. A missing dynamic property is defined to be the same as a dynamic property with value null. All dynamic properties are nullable.

On success, the service MUST respond with 204 No Content and an empty body.

Services MUST return an error if the property is not updatable.

Updating a primitive property or a complex property with a null value also sets the property to null.

11.4.9.3 Update a Complex Property

A successful PATCH request to the edit URL for a complex typed property updates that property. The request body MUST contain a single valid representation for the target complex type.

The service MUST directly modify only those properties of the complex type specified in the payload of the PATCH request.

The service MAY additionally support clients sending a PUT request to a URL that specifies a complex type. In this case, the service MUST replace the entire complex property with the values specified in the request body and set all unspecified properties to their default value.

Services MUST return an error if the property is not updatable.

11.4.9.4 Update a Collection Property

A successful PUT request to the edit URL of a collection property updates that collection. The message body MUST contain the desired new value, formatted as a collection property according to the specified format.

The service MUST replace the entire value with the value supplied in the request body.

A successful POST request to the edit URL of a collection property adds an item to the collection. The body of the request MUST be a single item to be added to the collection. If the collection is ordered, the item is added to the end of the collection, and $index MAY be used to specify a zero-based ordinal position to insert the new value, with a negative value indicating an ordinal position from the end of the collection.

A successful DELETE request to the edit URL of a collection property deletes all items in that collection.

Since collection members have no individual identity, PATCH is not supported for collection properties.

Services MUST return an error if the property is not updatable.

11.4.10 Managing Members of an Ordered Collection

Collections annotated with the Core.Ordered term (see [OData-VocCore]) have a stable order. Members of an ordered collection of primitive and complex types can be individually updated or deleted by invoking an update operation against the URL of the collection appended by a segment containing the zero-based ordinal of the item within the collection. A negative ordinal number indexes from the end of the collection, with -1 representing the last item in the collection.

Entities can be updated using their edit URL and SHOULD NOT be addressed using an index.

11.4.11 Positional Inserts

Collections of entity, complex, or primitive types annotated with the Core.PositionalInsert term (see [OData-VocCore]) support inserting items at a specific location via POST requests to the collection URL using the $index system query option. The value of the $index system query option is the zero-based ordinal position where the item is to be inserted. The ordinal positions of items within the collection greater than or equal to the inserted position are increased by one. A negative ordinal number indexes from the end of the collection, with -1 representing an insert as the last item in the collection.

Example 80: Insert a new email address at the second position

POST /service/Customers('ALFKI')/EmailAddresses?$index=1

Content-Type: application/json

{

"value": "alfred@futterkiste.de"

}

11.4.12 Update a Collection of Entities

Collections of entities can be updated by submitting a PATCH request to the resource path of the collection. The body of the request MUST be a delta payload, and the resource path of the collection MUST NOT contain type cast or filter segments, and MUST NOT contain any system query options that affect the shape of the result.

Added/changed entities are applied as upserts, and deleted entities as deletions. Non-key properties of deleted entities are ignored. The top-level collection may include added and deleted links, and related entities represented inline are updated according to the rules for treating related entities when updating an entity.

The response, if requested, is a delta payload, in the same structure and order as the request payload, representing the applied changes.

If the continue-on-error preference has been specified and any errors occur in processing the changes, then a delta response MUST be returned regardless of the return preference and MUST contain at least the failed changes. The service represents failed changes in the delta response as follows:

· Failed deletes in the request MUST be represented in the response as either entities or entity references, annotated with term Core.DataModificationException, see [OData-VocCore]. If the deleted entity specified a reason of deleted, the value of failedOperation MUST be delete, otherwise unlink.

· Failed inserts within the request MUST be represented in the response as deleted entities annotated with term Core.DataModificationException with a failedOperation value of insert.

· Failed updates within the request SHOULD be annotated in the response with term Core.DataModificationException with a failedOperation value of update.

· Failed added links within the request MUST represented in the response as deleted links annotated with term Core.DataModificationException with a failedOperation value of link.

· Failed deleted links within the request MUST represented in the response as added links annotated with term Core.DataModificationException with a failedOperation value of unlink.

· Collections within the request MUST be represented in the response as a collection with the current values and membership of the collection as it exists in the service after processing the request.

If an individual change fails due to a failed dependency, it MUST be annotated with term Core.DataModificationException and SHOULD specify a responseCode of 424 (Failed Dependency).

Alternatively, the verb PUT can be used, in which case the request body MUST be the representation of a collection of entities. In this case all entities provided in the request are applied as upserts, and any entities not provided in the request are deleted. In this case, if the continue-on-error preference has been specified, and the request returns a success response code, then a response MUST be returned regardless of the return preference, and MUST contain the full membership and values of the collection as it exists in the service.

If the continue-on-error preference has not been specified, and the service is unable to apply all of the changes in the request, then it MUST return an error response and MUST NOT apply any of the changes specified in the request payload.

11.4.13 Update Members of a Collection

Members of a collection can be updated by submitting a PATCH request to the URL constructed by appending /$each to the resource path of the collection. The additional path segment expresses that the request body describes an update to each member of the collection, not an update to the collection itself.

The resource path of the collection MAY contain type-cast or filter segments to subset the collection, see [OData‑URL].

For primitive-typed collections the body of the request MUST be a primitive value. Each member of the potentially filtered collection is updated to the specified primitive value.

For collections of structured type, the body of the request MUST be a full or partial representation of an instance of the collection’s structured type. Each member of the potentially filtered collection is updated using PATCH semantics. Structured types MAY include nested collections or delta collections, in which case the semantics described in Update a Collection of Entities applies.

Example 81: change the color of all beige-brown products

PATCH /service/Products/$filter(@bar)/$each?@bar=Color eq 'beige-brown'

Content-Type: application/json

{

"Color": "taupe"

}

The response, if requested, is a collection payload containing the updated representation of each member identified by the request. If the update payload includes nested collections or nested delta collections, then they MUST be included in the response, as described in Update a Collection of Entities.

Clients should note that requesting a response may be expensive for services that could otherwise efficiently apply updates to a (possibly filtered) collection.

If the continue-on-error preference has been specified, the service MAY continue processing updates after a failure. In this case, the service MUST return a response containing at least the members of the collection that failed to update, which MUST be annotated with term Core.DataModificationException with a failedOperation value of update.

If the continue-on-error preference has not been specified, and the service is unable to update all of the members identified by the request, then it MUST return an error response and MUST NOT apply any updates.

11.4.14 Delete Members of a Collection

Members of a collection can be deleted by submitting a DELETE request to the URL constructed by appending /$each to the resource path of the collection. The additional path segment expresses that the collection itself is not deleted.

The request resource path of the collection MAY contain type-cast or filter segments to subset the collection.

Example 82: delete all products older than 3

DELETE /service/Products/$filter(Age gt 3)/$each

If the path identifies a collection of entities and if the service returns a representation, then the response is a delta response containing a representation of a deleted entity for each deleted member.

If the collection is a collection of entities, then the client MAY specify the continue-on-error preference, in which case the service MAY continue processing deletes after a failure. In this case, the service MUST return a response containing at least an entity or entity reference for each entity identified by the request that failed to delete, which MUST be annotated with term Core.DataModificationException with a failedOperation value of delete.

Clients should note that requesting a response may be expensive for services that could otherwise efficiently apply deletes to a (possibly filtered) collection.

If the continue-on-error preference has not been specified, and the service is unable to delete all of the entities identified by the request, then it MUST return an error response and MUST NOT apply any changes.

11.5 Operations

Custom operations (Actions and Functions) allow encapsulating logic for modifying or requesting data that goes beyond simple CRUD described in the preceding sections of this chapter. See Action, ActionImport, Function, and FunctionImport in [OData-CSDLJSON] or [OData-CSDLXML].

11.5.1 Binding an Operation to a Resource

Actions and Functions MAY be bound to any type or collection, similar to defining a method in a class in object-oriented programming. The first parameter of a bound operation is the binding parameter.

The namespace- or alias-qualified name of a bound operation may be appended to any URL that identifies a resource whose type matches, or is derived from, the type of the binding parameter. The resource identified by that URL is used as the binding parameter value. Only aliases defined in the metadata document of the service can be used in URLs.

Example 83: the function MostRecentOrder can be bound to any URL that identifies a SampleModel.Customer

Example 84: invoke the MostRecentOrder function with the value of the binding parameter customer being the entity identified by http://host/service/Customers(6)

GET http://host/service/Customers(6)/SampleModel.MostRecentOrder()

Example 85: the function Comparison can be bound to any URL that identifies a collection of entities

Example 86: invoke the Comparison function on the set of red products

GET http://host/service/Products/$filter(Color eq 'Red')/Diff.Comparison()

11.5.2 Applying an Action to Members of a Collection

A bound operation with a single-valued binding parameter can be applied to each member of a collection by appending the path segment /$each to the resource path of the collection, followed by a forward slash and the namespace- or alias-qualified name of the bound operation. In this case the type of the collection members MUST match or be derived from the type of the binding parameter.

The resource path of the collection MAY contain type-cast or filter segments to subset the collection.

The response is a collection with members that are instances of the result type of the bound operation. If the bound operation returns a collection, the response is a collection of collections.

Example 87: invoke the MostRecentOrder function on each entity in the entity set Customers

GET http://host/service/Customers/$each/SampleModel.MostRecentOrder()

The client MAY specify the continue-on-error preference, in which case the service MAY continue processing actions after a failure. In this case, the service MUST, regardless of the return preference, return a response containing at least the members identified by the request for which the action failed. Such members MUST be annotated with term Core.DataModificationException with a failedOperation value of invoke.

If the continue-on-error preference has not been specified, and the service is unable to invoke the action against all of the entities identified by the request, then it MUST return an error response and MUST NOT apply the action to any of the members of the collection.

11.5.3 Advertising Available Operations within a Payload

Services MAY return actions and/or functions bound to a particular entity or entity collection as part of the representation of the entity or entity collection within the payload. The representation of an action or function depends on the format.

Example 88: given a GET request to http://host/service/Customers('ALFKI'), the service might respond with a Customer that includes the SampleEntities.MostRecentOrder function bound to the entity

{
"@context": ...,
"CustomerID": "ALFKI",
"CompanyName": "Alfreds Futterkiste",
  "#SampleEntities.MostRecentOrder": {
    "title": "Most Recent Order",
    "target": "Customers('ALFKI')/SampleEntities.MostRecentOrder()"
},
...
}

An efficient format that assumes client knowledge of metadata may omit actions and functions from the payload whose target URL can be computed via metadata following standard conventions defined in [OData‑URL].

Services can advertise that a function or action is not available for a particular instance by setting its value to null.

Example 89: the SampleEntities.MostRecentOrder function is not available for customer 'ALFKI'

{
"@context": ...,
"CustomerID": "ALFKI",
"CompanyName": "Alfreds Futterkiste",
"#SampleEntities.MostRecentOrder": null,
...
}

11.5.4 Functions

Functions are operations exposed by an OData service that MUST return data and MUST have no observable side effects.

11.5.4.1 Invoking a Function

To invoke a function bound to a resource, the client issues a GET request to a function URL. A function URL may be obtained from a previously returned entity representation or constructed by appending the namespace- or alias-qualified function name to a URL that identifies a resource whose type is the same as, or derived from, the type of the binding parameter of the function. The value for the binding parameter is the value of the resource identified by the URL prior to appending the function name, and additional parameter values are specified using inline parameter syntax. If the function URL is obtained from a previously returned entity representation, parameter aliases that are identical to the parameter name preceded by an at (@) sign MUST be used. Clients MUST check if the obtained URL already contains a query part and appropriately precede the parameters either with an ampersand (&) or a question mark (?).

Services MAY additionally support invoking functions using the unqualified function name by defining one or more default namespaces through the Core.DefaultNamespace term defined in [OData-VocCore].

Functions can be used within $filter or $orderby system query options. Such functions can be bound to a resource, as described above, or called directly by specifying the namespace- (or alias-) qualified function name. Parameter values for functions within $filter or $orderby are specified according to the inline parameter syntax.

To invoke a function through a function import the client issues a GET request to a URL identifying the function import and passing parameter values using inline parameter syntax. The canonical URL for a function import is the service root, followed by the name of the function import. Services MAY support omitting the parentheses when invoking a function import with no parameters, but for maximum interoperability MUST also support invoking the function import with empty parentheses.

If the function is composable, additional path segments may be appended to the URL that identifies the composable function (or function import) as appropriate for the type returned by the function (or function import). The last path segment determines the system query options and HTTP verbs that can be used with this this URL, e.g. if the last path segment is a multi-valued navigation property, a POST request may be used to create a new entity in the identified collection.

Example 90: add a new item to the list of items of the shopping cart returned by the composable MyShoppingCart function import

POST http://host/service/MyShoppingCart()/Items

...

Parameter values passed to functions MUST be specified either as a URL literal (for primitive values) or as a JSON formatted OData object (for complex values, or collections of primitive or complex values). Entity typed values are passed as JSON formatted entities that MAY include a subset of the properties, or just the entity reference, as appropriate to the function.

If a collection-valued function has no result for a given parameter value combination, the response is the format-specific representation of an empty collection. If a single-valued function with a nullable return-type has no result, the service returns 204 No Content.

If a single-valued function with a non-nullable return type has no result, the service returns 4xx. For functions that return a single entity 404 Not Found is the appropriate response code.

For a composable function the processing is stopped when the function result requires a 4xx response, and continues otherwise.

Function imports preceded by the $root literal MAY be used in the $filter or $orderby system query options, see [OData‑URL].

11.5.4.1.1 Inline Parameter Syntax

Parameter values are specified inline by appending a comma-separated list of parameter values, enclosed by parenthesis to the function name.

Each parameter value is represented as a name/value pair in the format Name=Value, where Name is the name of the parameter to the function and Value is the parameter value.

Example 91: invoke a Sales.EmployeesByManager function which takes a single ManagerID parameter via the function import EmployeesByManager

GET http://host/service/EmployeesByManager(ManagerID=3)

Example 92: return all Customers whose City property returns "Western" when passed to the Sales.SalesRegion function

GET http://host/service/Customers?
$filter=Sales.SalesRegion(City=$it/City) eq 'Western'

A parameter alias can be used in place of an inline parameter value. The value for the alias is specified as a separate query option using the name of the parameter alias.

Example 93: invoke a Sales.EmployeesByManager function via the function import EmployeesByManager, passing 3 for the ManagerID parameter

GET http://host/service/EmployeesByManager(ManagerID=@p1)?@p1=3

Services MAY in addition allow implicit parameter aliases for function imports and for functions that are the last path segment of the URL. An implicit parameter alias is the parameter name, optionally preceded by an at (@) sign. When using implicit parameter aliases, parentheses MUST NOT be appended to the function (import) name. The value for each parameter MUST be specified as a separate query option with the name of the parameter alias. If a parameter name is identical to a system query option name (without the optional $ prefix), the parameter name MUST be prefixed with an at (@) sign.

Example 94: invoke a Sales.EmployeesByManager function via the function import EmployeesByManager, passing 3 for the ManagerID parameter using the implicit parameter alias

GET http://host/service/EmployeesByManager?ManagerID=3

Non-binding parameters annotated with the term Core.OptionalParameter defined in [OData-VocCore] MAY be omitted. If it is annotated and the annotation specifies a DefaultValue, the omitted parameter is interpreted as having that default value. If omitted and the annotation does not specify a default value, the service is free on how to interpret the omitted parameter.

11.5.4.2 Function overload resolution

The same function name may be used multiple times within a schema, each with a different set of parameters. For unbound overloads the combination of the function name and the unordered set of parameter names MUST identify a particular function overload. For bound overloads the combination of the function name, the binding parameter type, and the unordered set of names of the non-binding parameters MUST identify a particular function overload.

All unbound overloads MUST have the same return type. Also, all bound overloads with a given binding parameter type MUST have the same return type.

If the function is bound and the binding parameter type is part of an inheritance hierarchy, the function overload is selected based on the type of the URL segment preceding the function name. A type-cast segment can be used to select a function defined on a particular type in the hierarchy, see [OData‑URL].

Non-binding parameters MAY be marked as optional by annotating them with the term Core.OptionalParameter defined in [OData-VocCore]. All parameters marked as optional MUST come after any parameters not marked as optional.

A function overload is selected if

· The set of specified parameters exactly matches a function overload, or else

· The set of specified parameters matches a subset of parameters that includes all non-optional parameters of exactly one function overload.

Services SHOULD avoid ambiguity, i.e. the combination of the function name, the unordered set of non-optional non-binding parameter names, plus the binding parameter type for bound overloads SHOULD identify a particular function overload. If there is ambiguity, then services MAY return 400 Bad Request with an error response body stating that the request was ambiguous.

11.5.5 Actions

Actions are operations exposed by an OData service that MAY have side effects when invoked. Actions MAY return data but MUST NOT be further composed with additional path segments.

11.5.5.1 Invoking an Action

To invoke an action bound to a resource, the client issues a POST request to an action URL. An action URL may be obtained from a previously returned entity representation or constructed by appending the namespace- or alias-qualified action name to a URL that identifies a resource whose type is the same as, or derives from, the type of the binding parameter of the action. The value for the binding parameter is the resource identified by the URL preceding the action name, and only the non-binding parameter values are passed in the request body according to the particular format.

Services MAY additionally support invoking actions using the unqualified action name by defining one or more default namespaces through the Core.DefaultNamespace term defined in [OData-VocCore].

To invoke an action through an action import, the client issues a POST request to a URL identifying the action import. The canonical URL for an action import is the service root, followed by the name of the action import. When invoking an action through an action import all parameter values MUST be passed in the request body according to the particular format.

Non-binding parameters that are nullable or annotated with the term Core.OptionalParameter defined in [OData-VocCore] MAY be omitted from the request body. If an omitted parameter is not annotated (and thus nullable), it MUST be interpreted as having the null value. If it is annotated and the annotation specifies a DefaultValue, the omitted parameter is interpreted as having that default value. If omitted and the annotation does not specify a default value, the service is free on how to interpret the omitted parameter. Note: a nullable non-binding parameter is equivalent to being annotated as optional with a default value of null.

4.01 services MUST support invoking actions with no non-binding parameters and parameterless action imports both without a request body and with a request body representing no parameters, according to the particular format. Interoperable clients SHOULD always include a request body, even when invoking actions with no non-binding parameters and parameterless action imports.

If the action returns results, the client SHOULD use content type negotiation to request the results in the desired format, otherwise the default content type will be used.

The client can request whether any results from the action be returned using the return Prefer header.

Actions that create and return a single entity follow the rules for entity creation and return a Location header that contains the edit URL or read URL of the created entity.

Actions without a return type respond with 204 No Content on success.

To request processing of the action only if the binding parameter value, an entity or collection of entities, is unmodified, the client includes the If-Match header with the latest known ETag value for the entity or collection of entities. The ETag value for a collection as a whole is transported in the ETag header of a collection response.

Example 95: invoke the SampleEntities.CreateOrder action using /Customers('ALFKI') as the customer (or binding parameter). The values 2 for the quantity parameter and BLACKFRIDAY for the discountCode parameter are passed in the body of the request. Invoke the action only if the customer’s ETag still matches.

POST http://host/service/Customers('ALFKI')/SampleEntities.CreateOrder

If-Match: W/"MjAxOS0wMy0yMVQxMzowNVo="

{
"items": [

{ "product": 4001, "quantity": 2 },
{ "product": 7062, "quantity": 1 }

],
"discountCode": "BLACKFRIDAY"
}

11.5.5.2 Action Overload Resolution

The same action name may be used multiple times within a schema provided there is at most one unbound overload, and each bound overload specifies a different binding parameter type.

If the action is bound and the binding parameter type is part of an inheritance hierarchy, the action overload is selected based on the type of the URL segment preceding the action name. A type-cast segment can be used to select an action defined on a particular type in the hierarchy, see [OData‑URL].

11.6 Asynchronous Requests

A Prefer header with a respond-async preference allows clients to request that the service process a Data Service Request asynchronously.

If the client has specified respond-async in the request, the service MAY process the request asynchronously and return a 202 Accepted response. A service MUST NOT reply to a Data Service Request with 202 Accepted if the request has not included the respond-async preference.

Responses that return 202 Accepted MUST include a Location header pointing to a status monitor resource that represents the current state of the asynchronous processing in addition to an optional Retry-After header indicating the time, in seconds, the client should wait before querying the service for status. Services MAY include a response body, for example, to provide additional status information.

A GET request to the status monitor resource again returns 202 Accepted response if the asynchronous processing has not finished. This response MUST again include a Location header and MAY include a Retry-After header to be used for a subsequent request. The Location header and optional Retry-After header may or may not contain the same values as returned by the previous request.

A GET request to the status monitor resource returns 200 OK once the asynchronous processing has completed. For OData 4.01 and greater responses, or OData 4.0 requests that include an Accept header that does not specify application/http, the response MUST include the AsyncResult response header. Any other headers, along with the response body, represent the result of the completed asynchronous operation. If the GET request to the status monitor includes an OData-MaxVersion header with a value of 4.0 and no Accept header, or an Accept header that includes application/http, then the body of the final 200 OK response MUST be represented as an HTTP message, as described in [RFC7230], which is the full HTTP response to the completed asynchronous operation.

A DELETE request sent to the status monitor resource requests that the asynchronous processing be canceled. A 200 OK or a 204 No Content response indicates that the asynchronous processing has been successfully canceled. A client can request that the DELETE should be executed asynchronously. A 202 Accepted response indicates that the cancellation is being processed asynchronously; the client can use the returned Location header (which MUST be different from the status monitor resource of the initial request) to query for the status of the cancellation. If a delete request is not supported by the service, the service returns 405 Method Not Allowed.

After a successful DELETE request against the status monitor resource, any subsequent GET requests for the same status monitor resource returns 404 Not Found.

If an asynchronous request is cancelled for reasons other than the consumers issuing a DELETE request against the status monitor resource, a GET request to the status monitor resource returns 200 OK with a response body containing a single HTTP response with a status code in the 5xx Server Error range indicating that the operation was cancelled.

The service MUST ensure that no observable change has occurred as a result of a canceled request.

If the client waits too long to request the result of the asynchronous processing, the service responds with a 410 Gone or 404 Not Found.

The status monitor resource URL MUST differ from any other resource URL.

11.7 Batch Requests

Batch requests allow grouping multiple individual requests into a single HTTP request payload. An individual request in the context of a batch request is a Metadata Request, Data Request, Data Modification Request, Action invocation request, or Function invocation request.

Batch requests are submitted as a single HTTP POST request to the batch endpoint of a service, located at the URL $batch relative to the service root.

Individual requests within a batch request are evaluated according to the same semantics used when the request appears outside the context of a batch request.

A batch request is represented using either the multipart batch format defined in this document or the JSON batch format defined in [OData-JSON].

11.7.1 Batch Request Headers

A batch request using the multipart batch format MUST contain a Content-Type header specifying a content type of multipart/mixed and a boundary parameter as defined in [RFC2046].

Example 96: multipart batch request

POST /service/$batch HTTP/1.1
Host: odata.org
OData-Version: 4.0
Content-Type: multipart/mixed; boundary=batch_36522ad7-fc75-4b56-8c71-56071383e77b

<Multipart Batch request body>

A batch request using the JSON batch format MUST contain a Content-Type header specifying a content type of application/json.

Example 97: JSON batch request

POST /service/$batch HTTP/1.1
Host: odata.org
OData-Version: 4.01
Content-Type: application/json

<JSON Batch request body>

Batch requests SHOULD contain the applicable OData-Version header.

Batch requests SHOULD contain an Accept header specifying the desired batch response format, either multipart/mixed or application/json. If no Accept header is provided, services SHOULD respond with the content type of the request.

If the set of request headers of a batch request are valid the service MUST return a 200 OK HTTP response code to indicate that the batch request was accepted for processing, but the processing is yet to be completed. The individual requests within the body of the batch request may subsequently fail or be malformed; however, this enables batch implementations to stream the results.

If the service receives a batch request with an invalid set of headers it MUST return a 4xx response code and perform no further processing of the batch request.

11.7.2 Request Dependencies

Requests within a batch may have dependencies on other requests according to the particular batch format.

In the JSON format, requests may explicitly declare a dependency on other requests that must be successfully processed before the current request. In addition, requests may be specified as part of an atomicity group whose members MUST either all succeed, or all fail. If a request fails, then any dependent requests within the JSON format return 424 Failed Dependency.

In the Multipart format, data modification requests or action invocation requests may be grouped as part of an atomic change set. Operations outside the change set are executed sequentially, while operations within the change set may be executed in any order.

11.7.3 Identifying Individual Requests

Each individual request within a batch request MAY have a request identifier assigned. The request identifier is case-sensitive, MUST be unique within the batch request, and MUST satisfy the rule request-id in [OData-ABNF].

The representation of the request identifier is format-specific, as are the rules for which individual requests require an identifier.

11.7.4 Referencing Returned Entities

Entities created by an insert request can be referenced in the request URL of subsequent requests by using the request identifier prefixed with a $ character as the first segment of the request URL.

If the $-prefixed request identifier is identical to the name of a top-level system resource ($batch, $crossjoin, $all, $entity, $root, $id, $metadata, or other system resources defined according to the OData-Version of the protocol specified in the request), then the reference to the top-level system resource is used. This collision can be avoided by e.g. using only numeric request identifiers.

Services MAY also support referencing within request bodies, in which case they SHOULD advertise this support by specifying the ReferencesInRequestBodiesSupported property in the Capabilities.BatchSupport term applied to the entity container, see [OData-VocCap].

11.7.5 Referencing the ETag of an Entity

Services MAY support the use of an ETag returned from a previous operation in an If-Match or If-None-Match header of a subsequent statement. Services SHOULD advertise this support by specifying the EtagReferencesSupported property in the Capabilities.BatchSupport annotation term applied to the entity container, see [OData-VocCap].

The ETag for a previous operation can be referenced by using the request identifier prefixed with a $ character as the unquoted value of the If-Match or If-None-Match header.

11.7.6 Referencing Values from Response Bodies

Services MAY support using values from a response body in the query part of the URL or in the request body of subsequent requests. Value references consist of a $ character, followed by the request identifier of the preceding request, and optionally followed by a valid OData path.

If the $-prefixed request identifier is identical to the name of a predefined literal for query expressions ($it, $root, or other literals defined according to the OData-Version of the protocol specified in the request), then the predefined literal is used. This collision can be avoided by e.g. using only numeric request identifiers.

11.7.7 Multipart Batch Format

The multipart batch format is represented as a Multipart Media Type message [RFC2046], a standard format allowing the representation of multiple parts, each of which may have a different content type.

11.7.7.1 Multipart Batch Request Body

The body of a multipart batch request is made up of a series of individual requests and change sets, each represented as a distinct body part (i.e. preceded by a boundary delimiter line consisting of two dashes and the value of the boundary parameter specified in the Content-Type header, and the last body part followed by a closing boundary delimiter line consisting of two dashes, the boundary, and another two dashes).

A body part representing an individual request MUST include a Content-Type header with value application/http.

The contents of a body part representing a change set MUST itself be a multipart document (see [RFC2046]) with one body part for each operation in the change set. Each body part representing an operation in the change set MUST specify a Content-ID header with a request identifier that is unique within the batch request.

A Content-Transfer-Encoding header with value binary may be included for historic reasons although this header is not used by HTTP and only needed for transmission via E-Mail. Neither clients nor services should rely on this header being present.

Preambles and epilogues in the multipart batch request body, as defined in [RFC2046], are valid but are assigned no meaning and thus MUST be ignored by processors of multipart batch requests.

The request URL of individual requests within a batch request or change set can use one of the following three formats:

Absolute URI with schema, host, port, and absolute resource path.

Example 98:

GET https://host:1234/path/service/People(1) HTTP/1.1

Absolute resource path and separate Host header

Example 99:

GET /path/service/People(1) HTTP/1.1

Host: myserver.mydomain.org:1234

Resource path relative to the batch request URI.

Example 100:

GET People(1) HTTP/1.1

Services MUST support all three formats for URLs of individual requests.

URLs must be correctly percent-encoded. For relative URLs this means that colons in the path part, especially within key values, MUST be percent-encoded to avoid confusion with the scheme separator. Colons within the query part, i.e. after the question mark character (?), need not be percent-encoded.

Each body part that represents a single request MUST NOT include:

authentication or authorization related HTTP headers
Expect, From, Max-Forwards, Range, or TE headers

Processors of batch requests MAY choose to disallow additional HTTP constructs in HTTP requests serialized within body parts. For example, a processor may choose to disallow chunked encoding to be used by such HTTP requests.

Example 101: a batch request that contains the following individual requests in the order listed

1. A query request

2. A change set that contains the following requests:

· Insert entity (with Content-ID = 1)

· Update entity (with Content-ID = 2)

3. A second query request

Note: For brevity, in the example, request bodies are excluded in favor of English descriptions inside <> brackets and OData-Version headers are omitted.

Note also that the two empty lines after the Host header of the GET request are necessary: the first is part of the GET request header; the second is the empty body of the GET request, followed by a CRLF according to [RFC2046].

POST /service/$batch HTTP/1.1
Host: host
OData-Version: 4.0
Content-Type: multipart/mixed; boundary=batch_36522ad7-fc75-4b56-8c71-56071383e77b

Content-Length: ###

--batch_36522ad7-fc75-4b56-8c71-56071383e77b
Content-Type: application/http

GET /service/Customers('ALFKI')
Host: host

--batch_36522ad7-fc75-4b56-8c71-56071383e77b
Content-Type: multipart/mixed; boundary=changeset_77162fcd-b8da-41ac-a9f8-9357efbbd

--changeset_77162fcd-b8da-41ac-a9f8-9357efbbd
Content-Type: application/http
Content-ID: 1

POST /service/Customers HTTP/1.1
Host: host
Content-Type: application/json
Content-Length: ###

<JSON representation of a new Customer>

--changeset_77162fcd-b8da-41ac-a9f8-9357efbbd
Content-Type: application/http
Content-ID: 2

PATCH /service/Customers('ALFKI') HTTP/1.1
Host: host
Content-Type: application/json
If-Match: xxxxx

Prefer: return=minimal
Content-Length: ###

<JSON representation of changes to Customer ALFKI>

--changeset_77162fcd-b8da-41ac-a9f8-9357efbbd--
--batch_36522ad7-fc75-4b56-8c71-56071383e77b
Content-Type: application/http

GET /service/Products HTTP/1.1
Host: host

--batch_36522ad7-fc75-4b56-8c71-56071383e77b--

11.7.7.2 Referencing New Entities

Entities created by an Insert request can be referenced in the request URL of subsequent requests within the same change set. Services MAY also support referencing across change sets, in which case they SHOULD advertise this support by specifying the ReferencesAcrossChangeSetsSupported property in the Capabilities.BatchSupport term applied to the entity container, see [OData-VocCap].

Example 102: a batch request that contains the following operations in the order listed:

A change set that contains the following requests:

· Insert a new entity (with Content-ID = 1)

· Insert a second new entity (references request with Content-ID = 1)

POST /service/$batch HTTP/1.1
Host: host
OData-Version: 4.0
Content-Type: multipart/mixed; boundary=batch_36522ad7-fc75-4b56-8c71-56071383e77b
Content-Length: ###

--batch_36522ad7-fc75-4b56-8c71-56071383e77b
Content-Type: multipart/mixed; boundary=changeset_77162fcd-b8da-41ac-a9f8-9357efbbd

--changeset_77162fcd-b8da-41ac-a9f8-9357efbbd
Content-Type: application/http
Content-ID: 1

POST /service/Customers HTTP/1.1
Host: host
Content-Type: application/json
Content-Length: ###

<JSON representation of a new Customer entity>
--changeset_77162fcd-b8da-41ac-a9f8-9357efbbd
Content-Type: application/http
Content-ID: 2

POST $1/Orders HTTP/1.1
Host: host
Content-Type: application/json
Content-Length: ###

<JSON representation of a new Order>
--changeset_77162fcd-b8da-41ac-a9f8-9357efbbd--
--batch_36522ad7-fc75-4b56-8c71-56071383e77b--

11.7.7.3 Referencing an ETag

Example 103: a batch request that contains the following operations in the order listed:

· Get an Employee (with Content-ID = 1)

· Update the salary only if the employee has not changed

POST /service/$batch HTTP/1.1
Host: host
OData-Version: 4.0
Content-Type: multipart/mixed; boundary=batch_36522ad7-fc75-4b56-8c71-56071383e77b
Content-Length: ###

--batch_36522ad7-fc75-4b56-8c71-56071383e77b
Content-Type: application/http
Content-ID: 1

GET /service/Employees(0) HTTP/1.1
Host: host
Accept: application/json

--batch_36522ad7-fc75-4b56-8c71-56071383e77b
Content-Type: application/http
Content-ID: 2

PATCH /service/Employees(0) HTTP/1.1
Host: host
Content-Type: application/json
Content-Length: ###
If-Match: $1

{

"Salary": 75000

}

11.7.7.4 Processing a Multipart Batch Request

The service MUST process the individual requests and change sets within a multipart batch request in the order received. Processing stops on the first error unless the continue-on-error preference is specified with an explicit or implicit value of true.

All requests in a change set represent a single change unit so a service MUST successfully process and apply all the requests in the change set or else apply none of them. It is up to the service implementation to define rollback semantics to undo any requests within a change set that may have been applied before another request in that same change set failed and thereby apply this all-or-nothing requirement. The service MAY execute the requests within a change set in any order and MAY return the responses to the individual requests in any order. If a request specifies a request identifier, the service MUST include the Content-ID header with the request identifier in the corresponding response so clients can correlate requests and responses.

11.7.7.5 Multipart Batch Response

A multipart response to a batch request MUST contain a Content-Type header with value multipart/mixed.

The body of a multipart response to a multipart batch request MUST structurally match one-to-one with the multipart batch request body, such that the same multipart message structure defined for requests is used for responses. There are three exceptions to this rule:

· When a request within a change set fails, the change set response is not represented using the multipart/mixed media type. Instead, a single response, using the application/http media type, is returned that applies to all requests in the change set and MUST be a valid OData error response.

· When an error occurs processing a request and the continue-on-error preference is not specified, or specified with an explicit value of false, processing of the batch is terminated and the error response is the last part of the multipart response.

· Asynchronously processed batch requests can return interim results and end with a 202 Accepted as the last part of the multipart response. Therefore, the respond-async preference MUST NOT be applied to individual requests within a batch if the batch response is a multipart response.

The body of a multipart response to a JSON batch request contains one body part for each processed or accepted request. The order of the body parts is insignificant as each body part MUST contain the Content-ID header with the value of the id name/value pair of the corresponding request object.

A response to an operation in a batch MUST be formatted exactly as it would have appeared outside of a batch as described in the corresponding subsections of chapter Data Service Requests. Relative URLs in each individual response are relative to the request URL of the corresponding individual request. URLs in responses MUST NOT contain $-prefixed request identifiers.

Example 104: referencing the batch request example 101 above, assume all the requests except the final query request succeed. In this case the response would be

HTTP/1.1 200 Ok
OData-Version: 4.0
Content-Length: ####
Content-Type: multipart/mixed; boundary=b_243234_25424_ef_892u748

--b_243234_25424_ef_892u748
Content-Type: application/http

HTTP/1.1 200 Ok
Content-Type: application/json
Content-Length: ###

<JSON representation of the Customer entity with key ALFKI>

--b_243234_25424_ef_892u748
Content-Type: multipart/mixed; boundary=cs_12u7hdkin252452345eknd_383673037

--cs_12u7hdkin252452345eknd_383673037
Content-Type: application/http
Content-ID: 1

HTTP/1.1 201 Created
Content-Type: application/json
Location: http://host/service.svc/Customer('POIUY')
Content-Length: ###

--cs_12u7hdkin252452345eknd_383673037
Content-Type: application/http
Content-ID: 2

HTTP/1.1 204 No Content
Host: host

--cs_12u7hdkin252452345eknd_383673037--

--b_243234_25424_ef_892u748
Content-Type: application/http

HTTP/1.1 404 Not Found
Content-Type: application/xml
Content-Length: ###

<Error message>
--b_243234_25424_ef_892u748—

11.7.7.6 Asynchronous Batch Requests

Batch requests MAY be executed asynchronously by including the respond-async preference in the Prefer header. If the service responds with a multipart batch response, it MUST ignore the respond-async preference for individual requests within a batch.

After successful execution of the batch request the response to the batch request is returned in the body of a response to an interrogation request against the status monitor resource URL (see Asynchronous Requests).

A service MAY return interim results to an asynchronously executing batch. It does this by responding with 200 OK to a GET request to the monitor resource and including a 202 Accepted response as the last part of the multipart response. The client can use the monitor URL returned in this 202 Accepted response to continue processing the batch response.

Since a change set is executed atomically, 202 Accepted MUST NOT be returned within a change set.

Example 105: referencing the example 101 above again, assume that

HTTP/1.1 202 Accepted
Location: http://service-root/async-monitor-0
Retry-After: ###

When interrogating the monitor URL only the first request in the batch has finished processing and all the remaining requests are still being processed. Note that the actual multipart batch response itself is contained in an application/http wrapper as it is a response to a status monitor resource:

HTTP/1.1 200 Ok

Content-Type: application/http

HTTP/1.1 200 Ok
OData-Version: 4.0
Content-Length: ####
Content-Type: multipart/mixed; boundary=b_243234_25424_ef_892u748

--b_243234_25424_ef_892u748
Content-Type: application/http

HTTP/1.1 200 Ok
Content-Type: application/json
Content-Length: ###

<JSON representation of the Customer entity with key ALFKI>

--b_243234_25424_ef_892u748
Content-Type: application/http

HTTP/1.1 202 Accepted
Location: http://service-root/async-monitor
Retry-After: ###

--b_243234_25424_ef_892u748--

After some time the client makes a second request using the returned monitor URL, not explicitly accepting application/http. The batch is completely processed and the response is the final result.

HTTP/1.1 200 Ok
AsyncResult: 200
OData-Version: 4.0

Content-Length: ####
Content-Type: multipart/mixed; boundary=b_243234_25424_ef_892u748

--b_243234_25424_ef_892u748
Content-Type: multipart/mixed; boundary=cs_12u7hdkin252452345eknd_383673037

--cs_12u7hdkin252452345eknd_383673037
Content-Type: application/http
Content-ID: 1

HTTP/1.1 201 Created
Content-Type: application/json
Location: http://host/service.svc/Customer('POIUY')
Content-Length: ###

<JSON representation of a new Customer entity>
--cs_12u7hdkin252452345eknd_383673037
Content-Type: application/http
Content-ID: 2

HTTP/1.1 204 No Content
Host: host

--cs_12u7hdkin252452345eknd_383673037--

--b_243234_25424_ef_892u748
Content-Type: application/http

HTTP/1.1 404 Not Found
Content-Type: application/xml
Content-Length: ###

<Error message>
--b_243234_25424_ef_892u748—

12 Security Considerations

This section is provided as a service to the application developers, information providers, and users of OData version 4.0 giving some references to starting points for securing OData services as specified. OData is a REST-full multi-format service that depends on other services and thus inherits both sides of the coin, security enhancements and concerns alike from the latter.

For HTTP relevant security implications please cf. the relevant sections of [RFC7231] (9. Security Considerations) and for the HTTP PATCH method [RFC5789] (5. Security Considerations) as starting points.

12.1 Authentication

OData Services requiring authentication SHOULD consider supporting basic authentication as defined in [RFC7617] over HTTPS for the highest level of interoperability with generic clients. They MAY support other authentication methods.

13 Conformance

OData is designed as a set of conventions that can be layered on top of existing standards to provide common representations for common functionality. Not all services will support all of the conventions defined in the protocol; services choose those conventions defined in OData as the representation to expose that functionality appropriate for their scenarios.

To aid in client/server interoperability, this specification defines multiple levels of conformance for an OData Service, as well as the minimal requirements for an OData Client to be interoperable across OData services.

13.1 OData 4.0 Service Conformance Levels

OData 4.0 defines three levels of conformance for an OData Service.

Note: The conformance levels are design to correspond to different service scenarios. For example, a service that publishes data compliant with one or more of the OData defined formats may comply with the OData 4.0 Minimal Conformance Level without supporting any additional functionality. A service that offers more control over the data that the client retrieves may comply with the OData 4.0 Intermediate Conformance Level. Services that conform to the OData 4.0 Advanced Conformance Level can expect to interoperate with the most functionality against the broadest range of generic clients.

Services can advertise their level of conformance by annotating their entity container with the term Capabilities.ConformanceLevel defined in [OData-VocCap].

Note: Services are encouraged to support as much additional functionality beyond their level of conformance as is appropriate for their intended scenario.

13.1.1 OData 4.0 Minimal Conformance Level

In order to conform to the OData 4.0 Minimal conformance level, a service:

1. MUST publish a service document at the service root (section 11.1.1)

2. MUST return data according to the[OData-JSON] format

3. MUST use server-driven paging when returning partial results (section 11.2.6.7) and not use any other mechanism

4. MUST return the appropriate OData-Version header (section 8.1.5)

5. MUST conform to the semantics the following headers, or fail the request

5.1. Accept (section 8.2.1)

5.2. OData-MaxVersion (section 8.2.7)

6. MUST follow OData guidelines for extensibility (section 6 and all subsections)

7. MUST successfully parse the request according to [OData-ABNF] for any supported system query options and either follow the specification or return 501 Not Implemented for any unsupported functionality (section 9.3.1)

8. MUST expose only data types defined in [OData-CSDLXML]

9. MUST NOT require clients to understand any metadata or instance annotations (section 6.4), custom headers (section 6.5), or custom content (section 6.2) in the payload in order to correctly consume the service

10. MUST NOT violate any OData update semantics (section 11.4 and all subsections)

11. MUST NOT violate any other OData-defined semantics

12. SHOULD support $expand (section 11.2.5.2)

13. SHOULD publish metadata at $metadata according to [OData-CSDLXML] and MAY publish metadata according to [OData-CSDLJSON] (section 11.1.2)

14. MUST support prefixed variants of supported headers and preference values

15. MUST support enumeration and duration literals in URLs with the type prefix

Additionally, if async operations are supported:

16. MUST return an HTTP message as the final response to an asynchronous request with an OData-MaxVersion value of 4.0 and an Accept header including application/http.

17. MAY return the AsyncResult header in the final response to an asynchronous request

To be considered an Updatable OData Service, the service additionally:

18. MUST include edit links (explicitly or implicitly) for all updatable or deletable resources according to [OData-JSON]

19. MUST support POST of new entities to insertable entity sets (section 11.4.1.5 and 11.4.2.1)

20. MUST support POST of new related entities to updatable navigation properties (section 11.4.2)

21. MUST support POST to $ref to add an existing entity to an updatable related collection (section 11.4.6.1)

22. MUST support PUT to $ref to set an existing single updatable related entity (section 11.4.6.3)

23. MUST support PATCH to all edit URLs for updatable resources (section 11.4.3)

24. MUST support DELETE to all edit URLs for deletable resources (section 11.4.5)

25. MUST support DELETE to $ref to remove a reference to an entity from an updatable navigation property (section 11.4.6.2)

26. MUST support If-Match header in update/delete of any resources returned with an ETag (section 11.4.1.1)

27. MUST return a Location header with the edit URL or read URL of a created resource (section 11.4.2)

28. MUST include the OData-EntityId header in response to any create or upsert operation that returns 204 No Content (section 8.3.4)

29. MUST support Upserts (section 11.4.4)

30. SHOULD support PUT and PATCH to an individual primitive (section 11.4.9.1) or complex (section 11.4.9.3) property (respectively)

31. SHOULD support DELETE to set an individual property to null (section 11.4.9.2)

32. SHOULD support deep inserts (section 11.4.2.2)

33. MAY support set-based updates (section 11.4.13) or deletes (section 11.4.14) to members of a collection

13.1.2 OData 4.0 Intermediate Conformance Level

In order to conform to the OData Intermediate Conformance Level, a service:

1. MUST conform to the OData 4.0 Minimal Conformance Level

2. MUST successfully parse the [OData-ABNF] and either follow the specification or return 501 Not Implemented for any unsupported functionality (section 9.3.1)

3. MUST support $select (section 11.2.5.1)

4. MUST support casting to a derived type according to [OData‑URL] if derived types are present in the model

5. MUST support $top (section 11.2.6.3)

6. MUST support /$value on media entities (section 4.12 in [OData‑URL]) and individual properties (section 11.2.4.1)

7. MUST support $filter (section 11.2.6.1)

7.1. MUST support eq, ne filter operations on properties of entities in the requested entity set (section 11.2.6.1.1)

7.2. MUST support aliases in $filter expressions (section 11.2.6.1.3)

7.3. SHOULD support additional filter operations (section 11.2.6.1.1) and MUST return 501 Not Implemented for any unsupported filter operations (section 9.3.1)

7.4. SHOULD support the canonical functions (section 11.2.6.1.2) and MUST return 501 Not Implemented for any unsupported canonical functions (section 9.3.1)

7.5. SHOULD support $filter on expanded entities (section 11.2.5.2.1)

8. SHOULD publish metadata at $metadata according to [OData-CSDLXML] (section 11.1.2)

9. SHOULD support the [OData-JSON] format

10. SHOULD consider supporting basic authentication as defined in [RFC7617] over HTTPS for the highest level of interoperability with generic clients

11. SHOULD support the $search system query option (section 11.2.6.6)

12. SHOULD support the $skip system query option (section 11.2.6.4)

13. SHOULD support the $count system query option (section 11.2.6.5)

14. SHOULD support $expand (section 11.2.5.2)

15. SHOULD support the lambda operators any and all on navigation- and collection-valued properties (section 5.1.1.10 in [OData‑URL])

16. SHOULD support the /$count segment on navigation and collection properties (section 11.2.10)

17. SHOULD support $orderby asc and desc on individual properties (section 11.2.6.2)

13.1.3 OData 4.0 Advanced Conformance Level

In order to conform to the OData Advanced Conformance Level, a service:

1. MUST conform to at least the OData 4.0 Intermediate Conformance Level

2. MUST publish metadata at $metadata according to [OData-CSDLXML] (section 11.1.2)

3. MUST support the [OData-JSON] format

4. MUST support the /$count segment on navigation and collection properties (section 11.2.10)

5. MUST support the lambda operators any and all on navigation- and collection-valued properties (section 5.1.1.10 in [OData‑URL])

6. MUST support the $skip system query option (section 11.2.6.4)

7. MUST support the $count system query option (section 11.2.6.5)

8. MUST support $orderby asc and desc on individual properties (section 11.2.6.2)

9. MUST support $expand (section 11.2.5.2)

9.1. MUST support returning references for expanded properties

9.2. MUST support $filter on expanded collection-valued properties

9.3. MUST support cast segment in expand with derived types

9.4. SHOULD support $orderby asc and desc on expanded collection-valued properties

9.5. SHOULD support $count on expanded collection-valued properties

9.6. SHOULD support $top and $skip on expanded collection-valued properties

9.7. SHOULD support $search on expanded collection-valued properties

9.8. SHOULD support $levels for recursive expand (section 11.2.5.2.1.1)

9.9. MAY support $compute on expanded properties

10. MUST support the $search system query option (section 11.2.6.6)

11. MUST support batch requests according to the multipart format (section 11.7 and all subsections) and MAY support batch requests according to the JSON Batch format defined in [OData-JSON]

12. MUST support the resource path conventions defined in [OData‑URL]

13. SHOULD support asynchronous requests (section 11.6)

14. SHOULD support Delta change tracking (section 11.3)

15. SHOULD support cross-join queries defined in [OData‑URL]

16. MAY support the $compute system query option (section 11.2.5.3)

13.2 OData 4.01 Service Conformance Levels

OData services can report conformance to the OData 4.01 specification by including 4.01 in the list of supported protocol versions in the Core.ODataVersions annotation, as defined in [OData-VocCore]. As all OData 4.01 compliant services must also be fully OData 4.0 compliant, OData 4.01 services do not need to separately list 4.0 as a supported version.

13.2.1 OData 4.01 Minimal Conformance Level

In order to conform to the OData 4.01 Minimal Conformance Level, a service:

1. MUST conform to the OData 4.0 Minimal Conformance Level

2. MUST be compliant with version 4.01 of the [OData-JSON] format

3. MUST return the AsyncResult result header in the final response to an asynchronous request if asynchronous operations are supported.

4. MUST support both prefixed and non-prefixed variants of supported headers and preference values

5. MUST reject a request with an incompatible $schemaversion system query option if a Core.SchemaVersion annotation is returned in $metadata

6. MUST support specifying supported system query options with or without the $ prefix

7. MUST support case-insensitive query option, operator, and canonical function names

8. MUST return identifiers in the case they are specified in $metadata

9. MUST support both 4.0 and 4.01 syntax in URLs for supported functionality regardless of requested OData-MaxVersion

a. MUST support casting strings to primitive types in URLs

b. MUST support enumeration and duration literals in URLs with or without the type prefix

c. MUST support invoking parameter-less function imports with or without parentheses

d. MUST support an empty object or no-content for the request body when invoking an action with no non-binding parameters

e. MUST support invoking functions and actions in a default namespace with or without namespace qualification

f. MUST support parameter aliases for key values and function parameter values if they allow the octets 00 (NUL), 2F (forward slash), or 5C (backslash) in string literals

g. SHOULD support implicit aliasing of parameters

h. SHOULD support eq/ne null comparison for navigation properties with a maximum cardinality of one

i. SHOULD support the in operator

j. SHOULD support divby

k. SHOULD support negative indexes for the substring function

l. MAY support Key-As-Segment URL convention

a. MUST also support canonical URL conventions (described in [OData‑URL]) or include URLs in payload

m. MAY support the count of a filtered collection in a common expression

n. MAY support equal and non-equal structural comparison

10. SHOULD publish metadata at $metadata according to both [OData-CSDLXML] and [OData-CSDLJSON] (section 11.1.2)

11. SHOULD NOT have identifiers within a uniqueness scope (e.g. a schema, a structural type, or an entity container) that differ only by case

12. SHOULD return the Core.ODataVersions annotation

13. SHOULD report capabilities through the Capabilities vocabulary

14. MAY support filtering on annotation values

15. MAY support $compute system query option

16. MAY support $search for all collections

17. MAY support 4.01 behavior, including returning 4.01 content and payloads, if the client does not specify the OData-MaxVersion:4.0 request header

In addition, to be considered an Updatable OData 4.01 Service, the service:

18. MUST conform to the OData 4.0 Minimal Conformance Level for an Updateable service.

19. MUST support DELETE to the reference of a collection member to be removed, identified by key (section 11.4.6.2)

20. SHOULD support PUT against single entity with nested content

21. SHOULD support deep updates (section 11.4.3.1) and deep inserts (section 11.4.2.2)

22. SHOULD support PUT or DELETE to $ref of a collection-valued nav prop

23. MAY support POST to collections of complex/primitive types

24. MAY support PATCH and DELETE to a collection

25. MAY support POST, PATCH and DELETE to a collection URL terminating in a type cast segment

26. MAY support PATCH to entity sets using the 4.01 delta payload format

27. MAY support $select and $expand on data modification requests

13.2.2 OData 4.01 Intermediate Conformance Level

In order to conform to the OData 4.01 Intermediate Conformance Level, a service:

1. MUST conform to the OData 4.01 Minimal Conformance Level

2. MUST conform to the OData 4.0 Intermediate Conformance Level

3. MUST support eq/ne null comparison for navigation properties with a maximum cardinality of one

4. MUST support the in operator

5. MUST support the $select option nested within $select

6. SHOULD support the count of a filtered collection in a common expression

7. SHOULD support equal and non-equal structural comparison

8. SHOULD support $compute system query option

9. SHOULD support nested query options in $select

10. MAY support nested parameter alias assignments in $select and $expand

11. MAY support filtering a collection using a /$filter path segment

13.2.3 OData 4.01 Advanced Conformance Level

In order to conform to the OData 4.01 Advanced Conformance Level, a service:

1. MUST conform to the OData 4.01 Intermediate Conformance Level

2. MUST conform to the OData 4.0 Advanced Conformance Level

3. MUST support the count of a filtered/searched collection in a common expression

4. MUST support $compute system query option

5. MUST support nested options in $select

5.1. MUST support $filter on selected collection-valued properties

5.2. SHOULD support $orderby asc and desc on selected collection-valued properties

5.3. SHOULD support the $count on selected collection-valued properties

5.4. SHOULD support $top and $skip on selected collection-valued properties

5.5. SHOULD support $search on selected collection-valued properties

6. MUST publish metadata at $metadata according to [OData-CSDLJSON] (section 11.1.2)

7. MUST support batch requests according both to the multipart format (section 11.7 and all subsections) and the JSON Batch format defined in [OData-JSON]

8. SHOULD support filtering a collection using a /$filter path segment

9. SHOULD support nested parameter alias assignments in $select and $expand

10. MAY support case-insensitive comparison of identifiers in URLs and request payloads if no exact match is found, using the same lookup sequence as for default namespaces with a case-insensitive comparison

13.3 Interoperable OData Clients

Interoperable OData clients can expect to work with OData Services that comply with at least the OData 4.0 Minimal Conformance Level and implement the [OData-JSON] format.

To be generally interoperable, OData clients

1. MUST specify the OData-MaxVersion header in requests (section 8.2.7)

2. MUST specify OData-Version (section 8.1.5) and Content-Type (section 8.1.1) in any request with a payload

3. MUST be a conforming consumer of OData as defined in [OData-JSON]

4. MUST follow redirects (section 9.1.5)

5. MUST correctly handle next links (section 11.2.6.7)

6. MUST support instances returning properties and navigation properties not specified in metadata (section 11.2)

7. MUST generate PATCH requests for updates, if the client supports updates (section 11.4.3)

8. MUST include the $ prefix when specifying OData-defined system query options

9. MUST use case-sensitive query options, operators, and canonical functions

10. SHOULD support basic authentication as defined in [RFC7617] over HTTPS

11. MAY request entity references in place of entities previously returned in the response (section 11.2.8)

12. MAY support deleted entities, link entities, deleted link entities in a delta response (section 11.3)

13. MAY support asynchronous responses (section 11.6)

14. MAY support metadata=minimal in a JSON response (see [OData-JSON])

15. MAY support streaming in a JSON response (see [OData-JSON])

In addition, interoperable OData 4.01 clients

16. MUST send OData 4.0-compliant payloads to services that don't advertise support for 4.01 or greater through the Core.ODataVersions metadata annotation (see [OData-VocCore])

17. MUST specify identifiers in payloads and URLs in the case they are specified in $metadata

18. MUST be prepared to receive any valid 4.01 CSDL

19. MUST be prepared to receive any valid 4.01 response according to the requested format

20. SHOULD use capabilities (see [OData-VocCap]) to determine if a 4.01 feature is supported but MAY attempt syntax and be prepared to handle either 501 Not Implemented or 400 Bad Request

Appendix A. Acknowledgments

The following individuals were members of the OASIS OData Technical Committee during the creation of this specification and their contributions are gratefully acknowledged:

· Howard Abrams (CA Technologies)

· Ken Baclawski (Northeastern University)

· Jay Balunas (Red Hat)

· Stephen Berard (Schneider Electric Industries SAS)

· Mark Biamonte (Progress Software)

· Matthew Borges (SAP SE)

· Edmond Bourne (BlackBerry)

· Joseph Boyle (Planetwork, Inc.)

· Peter Brown (Individual)

· Antonio Campanile (Bank of America)

· Pablo Castro (Microsoft)

· Axel Conrad (BlackBerry)

· Robin Cover (OASIS)

· Erik de Voogd (SDL)

· Yi Ding (Microsoft)

· Diane Downie (Citrix Systems)

· Patrick Durusau (Individual)

· Andrew Eisenberg (IBM)

· Chet Ensign (OASIS)

· Davina Erasmus (SDL)

· George Ericson (Dell)

· Colleen Evans (Microsoft)

· Jason Fam (IBM)

· Senaka Fernando (WSO2)

· Josh Gavant (Microsoft)

· Brent Gross (IBM)

· Zhun Guo (Individual)

· Anila Kumar GVN (CA Technologies)

· Stefan Hagen (Individual)

· Ralf Handl (SAP SE)

· Barbara Hartel (SAP SE)

· Hubert Heijkers (IBM)

· Jens Hüsken (SAP SE)

· Evan Ireland (SAP SE)

· Gershon Janssen (Individual)

· Ram Jeyaraman (Microsoft)

· Ted Jones (Red Hat)

· Diane Jordan (IBM)

· Stephan Klevenz (SAP SE)

· Gerald Krause (SAP SE)

· Nuno Linhares (SDL)

· Paul Lipton (CA Technologies)

· Susan Malaika (IBM)

· Ramanjaneyulu Malisetti (CA Technologies)

· Neil McEvoy (iFOSSF – International Free and Open Source Solutions Foundation)

· Stan Mitranic (CA Technologies)

· Dale Moberg (Axway Software)

· Graham Moore (BrightstarDB Ltd.)

· Farrukh Najmi (Individual)

· Shishir Pardikar (Citrix Systems)

· Sanjay Patil (SAP SE)

· Nuccio Piscopo (iFOSSF – International Free and Open Source Solutions Foundation)

· Michael Pizzo (Microsoft)

· Ramesh Reddy (Red Hat)

· Robert Richards (Mashery)

· Sumedha Rubasinghe (WSO2)

· James Snell (IBM)

· Jeffrey Turpin (Axway Software)

· John Willson (Individual)

· John Wilmes (Individual)

· Christopher Woodruff (Perficient, Inc.)

· Martin Zurmuehl (SAP SE)

Appendix B. Revision History

Revision	Date	Editor	Changes Made
Working Draft 01	2016-06-22	Michael Pizzo	Transferred content from OData 4.0 Part I - Protocol Errata 3
Committee Specification Draft 01	2016-12-08	Michael Pizzo Ralf Handl	Integrated 4.01 features · Schema Versioning · Preference omit-values · $compute system query option · Indexing into Ordered Collections · Deep Update · Improved referencing in batch requests
Committee Specification Draft 02	2017-06-20	Michael Pizzo Ralf Handl	More 4.01 features · Set-based operations · JSON Batch format · CSDL JSON format
Committee Specification Draft 03	2017-09-28	Michael Pizzo Ralf Handl	Incorporated review feedback Added nested query options for $select Added nested parameter alias value assignments within $expand and $select Simplified implicit parameter aliases for function (import) calls
Committee Specification Draft 04	2017-11-13	Michael Pizzo Ralf Handl	Incorporated review feedback Added deep insert of media entities Simplified unrelating entities Function imports in $filter and $orderby Stable order of action and function parameters
Committee Specification 01	2017-12-19	Michael Pizzo Ralf Handl	Non-Material Changes
Committee Specification Draft 05	2019-06-21	Michael Pizzo Ralf Handl	In-stream errors URL function matchesPattern Delta response for singletons Content-IDs in deep update Optional action and function parameters continue-on-error for operations on collections
Committee Specification Draft 06	2019-09-20	Michael Pizzo Ralf Handl	No dependencies to async requests within a batch request ETags in update responses Removed nested $expand within $select