- From: Klaus Weide <kweide@tezcat.com>
- Date: Wed, 3 Sep 1997 21:28:55 -0500 (CDT)
- To: Josh Cohen <josh@netscape.com>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
On Wed, 3 Sep 1997, Josh Cohen wrote:
> To: Foteos Macrides <MACRIDES@SCI.WFBR.EDU>
>
> > The "307" proposal will allow CGI scripts to work with at
> > least 99% probability of success with both HTTP/1.0 and HTTP/1.1
> > browsers. Think about it some more. Try it, you'll like it!!! :) :)
>
> So, am I correct in saying that the new CGI must check the version
> of the *REQUEST* before deciding which code to send back ?
Assuming we are talking about CGI scripts doing POST redirection with
current de-facto 302 behavior:
(1) Old scripts don't have to be rewritten (assuming that nothing else get
broken when they are transplanted into a HTTP/1.1 server environment.)
They can continue to use 302 (although it should be a simple change to
make them 2b instead).
(2) New scripts (your scenario) can
(a) act like old scripts, or
(b) check SERVER_PROTOCOL and use 303 or 302, depending, or
(c) use 303 to a HTTP/1.0 client if there's some out-of-band knowledge,
recognized user-agent, etc., or
(d) always use 303 even to all HTTP/1.0 clients.
(a) is for the lazy and the ignorant,
(b) is for those who want the best combination of unambiguousness [is that
a word?] and interoperability,
(c) is for the insiders,
(d) is for the adventurous and those who don't care.
So I think it's not absolutely necessary that a CGI script check the
client's version, but that it is a good idea.
Klaus
Received on Wednesday, 3 September 1997 19:31:48 UTC