httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Graham Leggett <>
Subject Re: cvs commit: httpd-2.0/modules/http http_protocol.c
Date Wed, 16 Apr 2003 08:24:07 GMT
William A. Rowe, Jr. wrote:

> -1 veto-not-vote.
> This breaks the protocol.

The patch is in response to a bug report which claims the existing 
behavior breaks the protocol.

So what is the correct behaviour supposed to be?

> Now, if you want to conditionally apply this logic, based either on the
> presence of a Via: header (meaning HTTP/1.1 carries different semantic
> meaning for the Date: header) or on the r->proxyreq flag, or both, that's fine.
> I'd also suggest parsing the date to assure it is legal before we substitute
> the value from the Date: response headers.

The proxy parses the Date header and converts it to canonical form 
before passing it on - this is moot though, as the Apache core throws 
this Date away and replaces it with its own.

> F.W.I.W., this is the reason for R-T-C on the 2.0 branch.  We aren't being
> pig-headed, we are trying to avoid regressions in the stable server.
> Not beating up on you - we live and learn.


-----------------------------------------		"There's a moon
					over Bourbon Street

View raw message