PATCH Method for HTTP

This is an Internet Standards Track document.

This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 5741.

Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc5789.

Copyright (c) 2010 IETF Trust and the persons identified as the document authors. All rights reserved.

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.

This specification defines the new HTTP/1.1 [RFC2616] method, PATCH, which is used to apply partial modifications to a resource.

A new method is necessary to improve interoperability and prevent errors. The PUT method is already defined to overwrite a resource with a complete new body, and cannot be reused to do partial changes. Otherwise, proxies and caches, and even clients and servers, may get confused as to the result of the operation. POST is already used but without broad interoperability (for one, there is no standard way to discover patch format support). PATCH was mentioned in earlier HTTP specifications, but not completely defined.

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

Furthermore, this document uses the ABNF syntax defined in Section 2.1 of [RFC2616].

The PATCH method requests that a set of changes described in the request entity be applied to the resource identified by the Request-URI. The set of changes is represented in a format called a "patch document" identified by a media type. If the Request-URI does not point to an existing resource, the server MAY create a new resource, depending on the patch document type (whether it can logically modify a null resource) and permissions, etc.

The difference between the PUT and PATCH requests is reflected in the way the server processes the enclosed entity to modify the resource identified by the Request-URI. In a PUT request, the enclosed entity is considered to be a modified version of the resource stored on the origin server, and the client is requesting that the stored version be replaced. With PATCH, however, the enclosed entity contains a set of instructions describing how a resource currently residing on the origin server should be modified to produce a new version. The PATCH method affects the resource identified by the Request-URI, and it also MAY have side effects on other resources; i.e., new resources may be created, or existing ones modified, by the application of a PATCH.

PATCH is neither safe nor idempotent as defined by [RFC2616], Section 9.1.

A PATCH request can be issued in such a way as to be idempotent, which also helps prevent bad outcomes from collisions between two PATCH requests on the same resource in a similar time frame. Collisions from multiple PATCH requests may be more dangerous than PUT collisions because some patch formats need to operate from a known base-point or else they will corrupt the resource. Clients using this kind of patch application SHOULD use a conditional request such that the request will fail if the resource has been updated since the client last accessed the resource. For example, the client can use a strong ETag [RFC2616] in an If-Match header on the PATCH request.

There are also cases where patch formats do not need to operate from a known base-point (e.g., appending text lines to log files, or non-colliding rows to database tables), in which case the same care in client requests is not needed.

The server MUST apply the entire set of changes atomically and never provide (e.g., in response to a GET during this operation) a partially modified representation. If the entire patch document cannot be successfully applied, then the server MUST NOT apply any of the changes. The determination of what constitutes a successful PATCH can vary depending on the patch document and the type of resource(s) being modified. For example, the common 'diff' utility can generate a patch document that applies to multiple files in a directory hierarchy. The atomicity requirement holds for all directly affected files. See "Error Handling", Section 2.2, for details on status codes and possible error conditions.

If the request passes through a cache and the Request-URI identifies one or more currently cached entities, those entries SHOULD be treated as stale. A response to this method is only cacheable if it contains explicit freshness information (such as an Expires header or "Cache-Control: max-age" directive) as well as the Content-Location header matching the Request-URI, indicating that the PATCH response body is a resource representation. A cached PATCH response can only be used to respond to subsequent GET and HEAD requests; it MUST NOT be used to respond to other methods (in particular, PATCH).

Note that entity-headers contained in the request apply only to the contained patch document and MUST NOT be applied to the resource being modified. Thus, a Content-Language header could be present on the request, but it would only mean (for whatever that's worth) that the patch document had a language. Servers SHOULD NOT store such headers except as trace information, and SHOULD NOT use such header values the same way they might be used on PUT requests. Therefore, this document does not specify a way to modify a document's Content-Type or Content-Language value through headers, though a mechanism could well be designed to achieve this goal through a patch document.

There is no guarantee that a resource can be modified with PATCH. Further, it is expected that different patch document formats will be appropriate for different types of resources and that no single format will be appropriate for all types of resources. Therefore, there is no single default patch document format that implementations are required to support. Servers MUST ensure that a received patch document is appropriate for the type of resource identified by the Request-URI.

Clients need to choose when to use PATCH rather than PUT. For example, if the patch document size is larger than the size of the new resource data that would be used in a PUT, then it might make sense to use PUT instead of PATCH. A comparison to POST is even more difficult, because POST is used in widely varying ways and can encompass PUT and PATCH-like operations if the server chooses. If the operation does not modify the resource identified by the Request-URI in a predictable way, POST should be considered instead of PATCH or PUT.

A server can advertise its support for the PATCH method by adding it to the listing of allowed methods in the "Allow" OPTIONS response header defined in HTTP/1.1. The PATCH method MAY appear in the "Allow" header even if the Accept-Patch header is absent, in which case the list of allowed patch documents is not advertised.

The security considerations for PATCH are nearly identical to the security considerations for PUT ([RFC2616], Section 9.6). These include authorizing requests (possibly through access control and/or authentication) and ensuring that data is not corrupted through transport errors or through accidental overwrites. Whatever mechanisms are used for PUT can be used for PATCH as well. The following considerations apply especially to PATCH.

A document that is patched might be more likely to be corrupted than a document that is overridden in entirety, but that concern can be addressed through the use of mechanisms such as conditional requests using ETags and the If-Match request header as described in Section 2. If a PATCH request fails, the client can issue a GET request to the resource to see what state it is in. In some cases, the client might be able to check the contents of the resource to see if the PATCH request can be resent, but in other cases, the attempt will just fail and/or a user will have to verify intent. In the case of a failure of the underlying transport channel, where a PATCH response is not received before the channel fails or some other timeout happens, the client might have to issue a GET request to see whether the request was applied. The client might want to ensure that the GET request bypasses caches using mechanisms described in HTTP specifications (see, for example, Section 13.1.6 of [RFC2616]).

Sometimes an HTTP intermediary might try to detect viruses being sent via HTTP by checking the body of the PUT/POST request or GET response. The PATCH method complicates such watch-keeping because neither the source document nor the patch document might be a virus, yet the result could be. This security consideration is not materially different from those already introduced by byte-range downloads, downloading patch documents, uploading zipped (compressed) files, and so on.

Individual patch documents will have their own specific security considerations that will likely vary depending on the types of resources being patched. The considerations for patched binary resources, for instance, will be different than those for patched XML documents. Servers MUST take adequate precautions to ensure that malicious clients cannot consume excessive server resources (e.g., CPU, disk I/O) through the client's use of PATCH.

PATCH is not a new concept, it first appeared in HTTP in drafts of version 1.1 written by Roy Fielding and Henrik Frystyk and also appears in Section 19.6.1.1 of RFC 2068.

Thanks to Adam Roach, Chris Sharp, Julian Reschke, Geoff Clemm, Scott Lawrence, Jeffrey Mogul, Roy Fielding, Greg Stein, Jim Luther, Alex Rousskov, Jamie Lokier, Joe Hildebrand, Mark Nottingham, Michael Balloni, Cyrus Daboo, Brian Carpenter, John Klensin, Eliot Lear, SM, and Bernie Hoeneisen for review and advice on this document. In particular, Julian Reschke did repeated reviews, made many useful suggestions, and was critical to the publication of this document.