httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From r.pl...@t-online.de (Rüdiger Plüm)
Subject Re: [PATCH 21492 30278 30419 31385] Several bugs in mod_cache / mod_disk_cache
Date Wed, 29 Sep 2004 11:24:29 GMT


Justin Erenkrantz wrote:
> --On Tuesday, September 28, 2004 5:28 PM +0200 Rüdiger Plüm 
> <r.pluem@t-online.de> wrote:

[..cut..]


Many thanks for the quick response. As I was curious I had a look to the cvs.
Sorry for the maybe stupid question. Do I understand this correctly: You commit the
changes to the main branch (aka. Apache 2.1) and Bill Stoddard backports them to the Apache
2.0 branch.
So the fixes for 21492 and 30278 will be part of 2.0.53 if Bill finds time to backport them
to the
the Apache 2.0 branch before 2.0.53 is released?

> 
>> 23687/30399 which is more an Enhancement than a bug.
> 
> 
> As for the patch that you do have, a better way to implement it is to 
> have a 3-state variable: UNSET|OFF|ON.  We use that idiom a lot.  It's 
> better than adding two variables.  Care to update your patch?

Yes, I will have a look into this. As I am a newbie to Apache programing I used the two
other FLAG directives (CacheIgnoreNoLastMod / CacheIgnoreCacheControl) as some kind of
template for my patch. They use two varaibles. Maybe it makes sense to migrate them also
to a 3-state variable. I will have a look into this.

> 
> Also, I'm wondering if we should extend this directive to be more 
> general: i.e. a 'CacheIgnoreHeader' directive?  What do you think?

Yes, this is a nice idea and it had been on my mind as I wrote the patch. I did not implement
that because

1. As mentioned above I am a newbie and the implementation of a FLAG directive seemed to be
easier
    to me than one that takes (multiple) parameters.
2. I am a little worried what happens if this directive is used to ignore headers that are
needed
    for the correct functionality of mod_cache (e.g. Etag / Content-Type). So maybe it is
needed to
    have a list of headers that cannot be ignored to prevent stupid configurations. But I
think
    this leads to the philosophical question how much flexibility should be given to the user
and
    how much the user should be protected from stupid configurations. So I am interested in
your thoughts
    on this point.

Nevertheless if the conclusion is made that creating a 'CacheIgnoreHeader' directive is a
good idea the next
question is how to setup this directive. I would think of something like the Options directive
that allows
a list of space separated arguments and allows to add or remove single arguments with +/-
during the merging
of arguments. Any thoughts?


> 
> I want to thank you *so* much for summarizing these bugs.  ;-)  I just 
> don't have the time to search through Bugzilla.  If you are aware of any 
> other mod_cache bugs, please bring them up on dev@httpd.  Thanks!  -- 

I will try to do so.

> justin
> 
> 

Regards

Rüdiger


Mime
View raw message