Network Working GroupJ. Reschke
Internet-Draftgreenbytes
Intended status: Standards TrackJuly 2009
Expires: January 2010

The Hypertext Transfer Protocol (HTTP) GET-Location header
draft-reschke-http-get-location-latest

Status of this Memo

This Internet-Draft is submitted to IETF in full conformance with the provisions of BCP 78 and BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts.

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as “work in progress”.

The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt.

The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html.

This Internet-Draft will expire in January 2010.

Copyright Notice

Copyright © 2009 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 in effect on the date of publication of this document (http://trustee.ietf.org/license-info). Please review these documents carefully, as they describe your rights and restrictions with respect to this document.

Abstract

Several hypertext transfer protocol (HTTP) extensions use methods other than GET to expose information. This has the drawback that this kind of information is harder to identify (missing a URL to which a GET request could be applied) and to cache.

This document specifies a simple extension header through which a server can advertise a substitute URL that an HTTP client subsequently can use with the GET method.

Editorial Note (To be removed by RFC Editor before publication)

Distribution of this document is unlimited. Please send comments to the Hypertext Transfer Protocol (HTTP) mailing list at ietf-http-wg@w3.org, which may be joined by sending a message with subject "subscribe" to ietf-http-wg-request@w3.org.

Discussions of the HTTP working group are archived at <http://lists.w3.org/Archives/Public/ietf-http-wg/>.

XML versions, latest edits and the issues list for this document are available from <http://greenbytes.de/tech/webdav/#draft-reschke-http-get-location>.


Table of Contents

Issues list

IdTypeStatusDateRaised By
content-locationeditopen2008-03-14julian.reschke@greenbytes.de
editeditopen2007-07-27julian.reschke@greenbytes.de
 I  edit   (type: edit, status: open)
julian.reschke@greenbytes.de2007-07-27 Umbrella issue for editorial fixes/enhancements.
Associated changes in this document: C.2, E.

1. Introduction

Several HTTP ([RFC2616]) extensions use methods other than GET to expose information. This has the drawback that this kind of information is harder to identify (missing a URL to which a GET request could be applied) and to cache.

This document specifies a simple extension header through which a server can advertise a substitute URL that an HTTP client subsequently can use with the GET method.

2. Notational Conventions

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].

The terminology used here follows and extends that in the HTTP specification [RFC2616].

3. The 'GET-Location' Header

The GET-Location entity header identifies a substitute resource that can be used in subsequent requests for the same information, but using the GET method.

Note that, by definition, the GET-Location header can only used on responses to safe methods.

Syntax (using the the augmented Backus-Naur Form (BNF) defined in Section 2.1 of [RFC2616]):

GET-Location = "GET-Location" ":" "<" Simple-ref ">"
                                  *( ";" location-directive ) )

location-directive = "etag=" entity-tag
                   | "max-age" "=" delta-seconds 
                   | location-extension

location-extension = token [ "=" ( token | quoted-string ) ]

Simple-ref     = absolute-URI | ( path-absolute [ "?" query ] )

absolute-URI   = <defined in [RFC3986], Section 4.3>
delta-seconds  = <defined in [RFC2616], Section 3.3.2>
entity-tag     = <defined in [RFC2616], Section 3.11>
path-absolute  = <defined in [RFC3986], Section 3.3>
quoted-string  = <defined in [RFC2616], Section 2.2>
query          = <defined in [RFC3986], Section 3.4>
token          = <defined in [RFC2616], Section 2.2>

Where:

Simple-ref
Contains either the URI or the absolute path of the location.
etag
The server can include the entity tag for the returned entity, if it would have been retrieved by a GET request to the substitute resource. Note that this is different from the value of the "ETag" header which could also be included in the response, but which would apply to the resource identified by the Request-URI.
max-age
Specifies a lifetime for the information returned by this header. A client MUST discard any information related to this header after the specified amount of time.

The freshness lifetime for the information obtained from the GET-Location header does not depend on the cacheability of the response it was obtained from (which, in general, may not be cacheable at all). The "max-age" directive allows the server to specify after how many seconds a client should discard knowledge about the alternate resource. In absence of that header, clients SHOULD discard the information after 3600 seconds.

There is no direct relation between the status code of the HTTP response that included GET-Location and the status codes for subsequent GET requests on the substitute resource. For instance, GET-Location could be included in a 207 response to PROPFIND ([RFC4918], Section 9.1), but the response code for a succesful GET on the substitute resource would usually be 200.

Note that servers may, but are not required to support methods other than GET or head on the substitute resource.

4. Security Considerations

This specification introduces no new security considerations beyond those discussed in Section 15 of [RFC2616].

5. IANA Considerations

This document specifies the new HTTP header listed below, to be added to the permanent registry (see [RFC3864]).

Header field name:
GET-Location
Applicable protocol:
http
Status:
standard
Author/Change controller:
IETF
Specification document:
Section 3 of this specification

6. Acknowledgments

This document has benefited from thoughtful discussion by Stefan Eissing and Henrik Nordstrom.

7. References

7.1 Normative References

[RFC2119]Bradner, S., “Key words for use in RFCs to Indicate Requirement Levels”, BCP 14, RFC 2119, March 1997.
[RFC2616]Fielding, R., Gettys, J., Mogul, J., Frystyk, H., Masinter, L., Leach, P., and T. Berners-Lee, “Hypertext Transfer Protocol -- HTTP/1.1”, RFC 2616, June 1999.
[RFC3986]Berners-Lee, T., Fielding, R., and L. Masinter, “Uniform Resource Identifier (URI): Generic Syntax”, RFC 3986, January 2005.

7.2 Informative References

[draft-gregorio-uritemplate]Gregorio, J., Ed., Hadley, M., Ed., Nottingham, M., Ed., and D. Orchard, “URI Template”, Internet-Draft draft-gregorio-uritemplate-03 (work in progress), March 2008.
[RFC2068]Fielding, R., Gettys, J., Mogul, J., Nielsen, H., and T. Berners-Lee, “Hypertext Transfer Protocol -- HTTP/1.1”, RFC 2068, January 1997.
[RFC3253]Clemm, G., Amsden, J., Ellison, T., Kaler, C., and J. Whitehead, “Versioning Extensions to WebDAV”, RFC 3253, March 2002.
[RFC3864]Klyne, G., Nottingham, M., and J. Mogul, “Registration Procedures for Message Header Fields”, BCP 90, RFC 3864, September 2004.
[RFC4918]Dusseault, L., Ed., “HTTP Extensions for Web Distributed Authoring and Versioning (WebDAV)”, RFC 4918, June 2007.

A. Examples

A.1 WebDAV Collection Membership

In this example the client uses the WebDAV PROPFIND method ("HTTP Extensions for Web Distributed Authoring and Versioning", [RFC4918], Section 9.1) to get a list of all collection members, along with their DAV:resourcetype property ([RFC4918], Section 15.9):

>>Request

PROPFIND /collection/ HTTP/1.1 
Host: example.com
Depth: 1
Content-Type: application/xml

<propfind xmlns="DAV:">
  <prop>
    <resourcetype/>
  </prop>
</propfind>

The response contains the requested information, plus the GET-Location header, identifying a separate resource which can provide the same information using the HTTP GET method:

>>Response

HTTP/1.1 207 Multi-Status
Content-Type: application/xml
GET-Location: <http://example.com/collection/;members>; etag="123";
              max-age=3600

<multistatus xmlns="DAV":>
  <response>
    <href>/collection/</href>
    <propstat>
      <prop>
        <resourcetype><collection/></resourcetype>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
  <response>
    <href>/collection/member</href>
    <propstat>
      <prop>
        <resourcetype/>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
</multistatus>

The response provided the URL of the substitute resource, so when the client wishes to refresh the collection information, it uses that URI. The response contained the entity tag for the data being returned, so it can make the request conditional:

>>Request

GET /collection/;members HTTP/1.1 
Host: example.com
Accept: application/xml
If-None-Match: "123"

The information did not change, so the server does not need to return new data:

>>Response

HTTP/1.1 304 Not Modified

Later on, the client tries again. This time, however, a second member has been added:

>>Request

GET /collection/;members HTTP/1.1 
Host: example.com
Accept: application/xml
If-None-Match: "123"

>>Response

HTTP/1.1 200 OK
Content-Type: application/xml
ETag: "124"

<multistatus xmlns="DAV":>
  <response>
    <href>/collection/</href>
    <propstat>
      <prop>
        <resourcetype><collection/></resourcetype>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
  <response>
    <href>/collection/member</href>
    <propstat>
      <prop>
        <resourcetype/>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
  <response>
    <href>/collection/member2</href>
    <propstat>
      <prop>
        <resourcetype/>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
</multistatus>

Finally, the collection has been removed by somebody else. The client tries a refresh:

>>Request

GET /collection/;members HTTP/1.1 
Host: example.com
Accept: application/xml
If-None-Match: "124"

>>Response

HTTP/1.1 404 Not Found

Note that it may be hard to compute entity tags for more complex PROPFIND responses. For instance, most properties depend on the state of the collection member, not the state of the collection itself, and thus the response will change even though the state of the collection itself did not change.

This is why this extension leaves it to the server whether to return a GET-Location at all, and if so, whether to return cache validators along with it.

A.2 WebDAV Custom Properties

Here, the client uses the WebDAV PROPFIND method ([RFC4918], Section 9.1) to obtain a custom property:

>>Request

PROPFIND /collection/member HTTP/1.1 
Host: example.com
Depth: 0
Content-Type: application/xml

<propfind xmlns="DAV:">
  <prop>
    <title xmlns="http://ns.example.com/"/>
  </prop>
</propfind>

>>Response

HTTP/1.1 207 Multi-Status
Content-Type: application/xml
GET-Location: </collection/member;prop=title>; etag="1"

<multistatus xmlns="DAV":>
  <response>
    <href>/collection/member</href>
    <propstat>
      <prop>
        <title xmlns="http://ns.example.com/"
        >Document Title</title>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
</multistatus>

>>Request

GET /collection/member;prop=title HTTP/1.1 
Host: example.com
If-None-Match: "1"

>>Response

HTTP/1.1 304 Not Modified

Later, the request is repeated after the title property indeed changed...:

>>Request

GET /collection/member;prop=title HTTP/1.1 
Host: example.com
If-None-Match: "1"

>>Response

HTTP/1.1 200 OK
Content-Type: application/xml
ETag: "2"

<multistatus xmlns="DAV":>
  <response>
    <href>/collection/member</href>
    <propstat>
      <prop>
        <title xmlns="http://ns.example.com/"
        >New Document Title</title>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
</multistatus>

Although this example may look like every WebDAV property would need a separate entity tag, this is of course not the case. For instance, a server that stores all custom properties in a single place (like a properties file) could use the same computation for the entity tag for all properties. Also, it could implement resources representing multiple custom property values the same way.

A.3 DeltaV Version History Report

Here, the client uses the DeltaV DAV:version-tree report ("Versioning Extensions to WebDAV", [RFC3253], Section 3.7) to obtain the members of the version history of a version-controlled resource.

>>Request

REPORT /collection/member HTTP/1.1 
Host: example.com
Depth: 0
Content-Type: application/xml

<version-tree xmlns="DAV:">
  <prop>
    <resourcetype/>
  </prop>  
</version-tree>

>>Response

HTTP/1.1 207 Multi-Status
Content-Type: application/xml
GET-Location: </version-storage/12345/;justmembers>

<multistatus xmlns="DAV":>
  <response>
    <href>/version-storage/12345/V1</href>
    <propstat>
      <prop>
        <resourcetype><collection/></resourcetype>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
  <response>
    <href>/version-storage/12345/V2</href>
    <propstat>
      <prop>
        <resourcetype><collection/></resourcetype>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
</multistatus>

Note that in this case, the substitute resource can be almost identical to the one from the PROPFIND/Depth:1 example: the only difference being that the report result does not contain a DAV:response element for the collection itself.

B. Related HTTP features

This section discusses some related HTTP features and explains why the author feels that they can't be used for the given use case.

B.1 Status 303

Section 10.3.4 of [RFC2616] defines the status code 303 (See Other):

The response to the request can be found under a different URI and SHOULD be retrieved using a GET method on that resource. This method exists primarily to allow the output of a POST-activated script to redirect the user agent to a selected resource. The new URI is not a substitute reference for the originally requested resource. The 303 response MUST NOT be cached, but the response to the second (redirected) request might be cacheable.

On first glance, it may look as if this addresses exactly the given use case. However:

  1. It says: "The new URI is not a substitute reference for the originally requested resource. The 303 response MUST NOT be cached, but the response to the second (redirected) request might be cacheable." That is, the information about the alternate resource is not cacheable.
  2. Servers returning a 303 status instead of the one expected by the client, such as 207 Multistatus, would likely break existing clients.

B.2 Content-Location Header

The Content-Location value is not a replacement for the original requested URI; it is only a statement of the location of the resource corresponding to this particular entity at the time of the request. (...)

However, the purpose of "GET-Location" is to enable the server to provide a permanent replacement URI.

 I  content-location   (type: edit, status: open)
julian.reschke@greenbytes.de2008-03-14 In http://lists.w3.org/Archives/Public/ietf-http-wg/2007JulSep/0184.html, Roy Fielding points out that the statement about Content-Location is out-of context; and that Content-Location indeed should be used for this use case.
In this case, other information defined in GET-Location would still be missing, such as validatity and etag information (see response in http://lists.w3.org/Archives/Public/ietf-http-wg/2007JulSep/0195.html).
So would it make sense to use Content-Location instead, and move the optional fields defined in GET-Location somewhere else? Or would Content-Location alone be sufficient (which proper instructions how to use it with methods like PROPFIND and REPORT)?

B.3 Location header

The Location response-header field is used to redirect the recipient to a location other than the Request-URI for completion of the request or identification of a new resource. (...)

Neither of these cases ("redirect to a location for completion of the request" and "identification of a new resource") matches the use case "GET-Location" covers.

C. Alternate Approaches

C.1 Link Relation

An alternative to introducing a new header would be to re-use an existing header, such as the Link header defined in Section 19.6.2 of [RFC2068]. Note that this still would require registering a link relation.

The example from Appendix A.1 would then read like this:

Link: </collection/;members>;
      rel=getlocation; etag="123"; max-age=3600

C.2 Multistatus Body Extension

Observing that the whole proposal tries to deal with WebDAV related shortcomings, it may make sense to constrain the solution to WebDAV response bodies, thereby not having to introduce anything that would be visible outside WebDAV.

A very simple approach would be to embed the information in the DAV:multistatus ([RFC4918], Section 14.16) response body.

Re-using the example in Appendix A.1, this could look like this:

HTTP/1.1 207 Multi-Status
Content-Type: application/xml

<multistatus xmlns="DAV":>
  <gl:get-location
      xmlns:gl=" I http://purl.oclc.org/NET/webdav/mount/getlocationhttp://purl.net/NET/webdav/getlocation">
    <href>/collection/;members</href>
    <getetag>"123"</getetag>
    <gl:max-age>3600</gl:max-age>
  <gl:get-location
  <response>
    <href>/collection/</href>
    <propstat>
      <prop>
        <resourcetype><collection/></resourcetype>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
  <response>
    <href>/collection/member</href>
    <propstat>
      <prop>
        <resourcetype/>
      </prop>
      <status>HTTP/1.1 200 OK</status>
    </propstat>
  </response>
</multistatus>

D. Open Issues

D.1 Content Negotiation on GET-Location

Should it be possible to use Content Negotiation on the resource identified by GET-Location? A use case could be a metadata provider that would support different formats, such as WebDAV's multistatus format (MIME type missing!), RDF, JSON, whatever.

This could be done using a location-extension specifying the Accept header for the GET operation.

D.2 Using URI Templates rather than URIs

Should we allow servers to return URI templates ([draft-gregorio-uritemplate]), so that clients can compute substitute URLs for other requests as well?

For instance, this could be done by allowing a URI template instead of the Simple-ref, and to return another template specifying how to derive the template variable from the Request-URI:

>>Request

PROPFIND /documents/a/b HTTP/1.1 
Host: example.com
Depth: 0
Content-Type: application/xml

>>Response

HTTP/1.1 207 Multi-Status
Content-Type: application/xml
GET-Location: </metadata/{path};members>; path-template=</a/b/{path}>

...

So in this case, the actual URI to be used would be <http://example.com/metadata/a/b;members>.

D.3 Extensions

Do we need a registry for new location-directive values?

E. Change Log (to be removed by RFC Editor before publication)

E.1 Since draft-reschke-http-get-location-00

Add and resolve issues "non-get" and "status-codes". Add issue "content-location". Add "Acknowledgments" Section. Update uri-template reference. Discuss more alternative approaches: Link header, Multistatus body extension.

 I  

E.2 Since draft-reschke-http-get-location-01

Author's Address

Julian F. Reschke
greenbytes GmbH
Hafenweg 16
Muenster, NW 48155
Germany
Phone: +49 251 2807760
EMail: julian.reschke@greenbytes.de
URI: http://greenbytes.de/tech/webdav/