You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Randy Terbush <ra...@zyzzyva.com> on 1997/06/25 07:04:04 UTC

mod_cern_meta problems with recent change

Seems that the recent change to mod_cern_meta has caused a problem.
All requests for documents using content negotiation seem to fail 
with the server complaining that it cannot read the .web or .meta 
file. I have not tracked this down beyond the fact that remove 
mod_cern_meta from the server solves the problem.





Re: mod_cern_meta problems with recent change

Posted by Dean Gaudet <dg...@arctic.org>.
I just committed a fix.  We can either go with that, or I can reinstate
the stat().  The stat check for mode == 0 is unneeded in any case (the
other mode checks against 0 are a hack that's used in the request
structure to indicate that the file exists and has a known mode).

BTW, there's an oddity with get_path_info().  If the request is for
/path/to/.web/foobar.meta, and /path/to exists and is a directory, but
/path/to/.web doesn't exist, then it sets path_info == "/foobar.meta" and
keeps the .web around on the filename.  Odd. 

And another BTW, these sub_req_lookup tests to avoid the security problems
are really heavy-handed.  But I can't think of anything else that covers
all the cases.

Dean

On Wed, 25 Jun 1997, Randy Terbush wrote:

> Seems that the recent change to mod_cern_meta has caused a problem.
> All requests for documents using content negotiation seem to fail 
> with the server complaining that it cannot read the .web or .meta 
> file. I have not tracked this down beyond the fact that remove 
> mod_cern_meta from the server solves the problem.
> 
> 
> 
> 
> 


Re: mod_cern_meta problems with recent change

Posted by Marc Slemko <ma...@worldgate.com>.
Do you have any access configs on your server to limit requests so that
*.web and *.meta files can't be read?

Dean changed it to do a subrequest to check that the metafile can be read
instead of just a stat.

(no, I haven't looked at it either...)

On Wed, 25 Jun 1997, Randy Terbush wrote:

> Seems that the recent change to mod_cern_meta has caused a problem.
> All requests for documents using content negotiation seem to fail 
> with the server complaining that it cannot read the .web or .meta 
> file. I have not tracked this down beyond the fact that remove 
> mod_cern_meta from the server solves the problem.
> 
> 
> 
>