W3C home > Mailing lists > Public > ietf-dav-versioning@w3.org > July to September 2001

RE: Cache-Control and the Vary header...

From: Clemm, Geoff <gclemm@rational.com>
Date: Thu, 6 Sep 2001 15:17:16 -0400
Message-ID: <3906C56A7BD1F54593344C05BD1374B1042AB457@SUS-MA1IT01>
To: ietf-dav-versioning@w3.org
   From: Peter Raymond [mailto:Peter.Raymond@merant.com]

   While reading section 8.3 of the draft I noticed that we explicitly
   specify that the HTTP Vary header must be sent on responses to
   methods that take Label headers.  I assume this is because when a
   user does a caching operation and specifies a Label he/she does not
   want the cached response to come back if they do the same method on
   the same URL but with a different Label header or without a Label
   header.

Yes.

   But thinking along the same lines why do we not include the Depth
   header in the Vary field?  A request on a URL can return different
   results depending on the presence or the value of the depth header.

The Depth header does not affect GET and HEAD (which are the main
cacheable requests).

   Also the text in section 8.3 is wrong, it says that GET and
   PROPFIND are cacheable requests but RFC2518 (section 8.1 for
   example) says that the methods MUST NOT be cached (eg
   Cache-Control: no-cache).

Good catch!  PROPFIND is not cacheable (but GET is ... section 8.1
of 2518 is just about PROPFIND).  I'll fix that by deleting the reference
to PROPFIND.

   If both GET and PROPFIND are not cacheable why have the reference
   to the Vary header at all?

The Label header affects GET (unlike the Depth header), which is why
the Vary header is needed.

Cheers,
Geoff
Received on Thursday, 6 September 2001 15:06:20 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 13:57:42 GMT