Re: GET/HEAD support "MUST"

• From: Julian Reschke <julian.reschke@gmx.de>
• Date: Thu, 29 Jan 2009 22:56:16 +0100
• To: "William A. Rowe, Jr." <wrowe@rowe-clan.net>
• CC: Geoffrey Sneddon <foolistbar@googlemail.com>, HTTP Working Group <ietf-http-wg@w3.org>
• Message-ID: <49822600.20003@gmx.de>

William A. Rowe, Jr. wrote:
>>> a) Although the server MUST support those two methods, MUST it support
>>> it for every resource?
>> Dunno.
> 
> It must support them in tandem.  Implementing HEAD that corresponds to
> anything except GET is not valid.  Implementing GET without HEAD is not
> valid either.

Right. This is so hardwired into my brain that I didn't even consider 
this being not clear. We probably should add a statement about that to 
the description of the GET method.

> Nothing says a resource must support them, however.  E.g. PUT to a specific
> URI might be valid, but there is no resource yet before PUT, ergo GET and
> HEAD would be 404 not found.

Indeed.

> It's probably a good idea to implement them, in a write-only server they
> could consistently return 404.

That's probably the safest approach.

BR, Julian

Received on Thursday, 29 January 2009 21:57:04 UTC