httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "William A. Rowe, Jr." <>
Subject Re: cvs commit: httpd-2.0/server request.c
Date Fri, 09 Nov 2001 07:49:45 GMT
From: "William A. Rowe, Jr." <>
Sent: Wednesday, October 17, 2001 9:24 AM

> From: "Rodent of Unusual Size" <Ken.Coar@Golux.Com>
> Sent: Wednesday, October 17, 2001 10:11 AM
> > "Roy T. Fielding" wrote:
> > > 
> > > Subrequests are file content inclusions, not internal redirects.
> > > They aren't supposed to check Location blocks because
> > > their URLs are file URLs internal to the server.
> > 
> > I'm still not caught up on all aspects of 2.0, but that certainly
> > doesn't jibe with 'include virtual=' on 1.3, which I believe was
> > handled with a subrequest..
> The patch I posted last night reverts to the 1.3 behavior, with one exception;
> sub_req_lookup_uri now perform <Location /> walks that were not performed
> in Apache 1.3.  They already called translate_name back in 1.3.
> Greg's and Roy's observation is that sub_req_lookup_file shouldn't perform
> either of the <Location /> walks or call translate_name.  In fact they are
> correct, those didn't occur in 1.3 either.  We simply need to call out that
> sub_req_lookup_file/dirent won't perform any Location features or Translation.

Important or not so important footnote.  We use sub_req_lookup_dirent, which is
a flavor of sub_req_lookup_file, to handle autoindex exclusions.  If the user
protects a <location >, that -file- can still appear in an autoindex listing.

I dunno if that concerns anyone.  Perhaps it should be called out in the docs.

> Note one bit of grand bogusness, however.  We used a very strange combination
> of r->main's per_dir_config (including it's Vhost/Directory/File/Location merged
> per-dir-configs, in that order) when constructing these subrequests.
> I scream uncle, what makes sense for file subrequests?  Use the parent request's
> Vhost+Location, then merge the correct Directory/File over that base?  I have no 
> good answer.

I'm still waiting ... nattering nabobs of negativism please revisit your assertions.


View raw message