Re: svn commit: r905406 - in /httpd/httpd/trunk: CHANGES docs/manual/mod/mod_include.xml modules/filters/mod_include.c

[Nick Kew <ni...@webthing.com>]

On Mon, 01 Feb 2010 20:43:01 -0000
minfrin@apache.org wrote:

> Author: minfrin
> Date: Mon Feb  1 20:43:01 2010
> New Revision: 905406
> 
> URL: http://svn.apache.org/viewvc?rev=905406&view=rev
> Log:
> Allow fine control over the removal of Last-Modified and ETag headers
> within the INCLUDES filter, making it possible to cache responses if
> desired. Fix the default value of the SSIAccessEnable directive.

Interesting.

However, given what you've implemented, I think it would perhaps
benefit from a third state for ETag and Last-Modified.  Namely,
allow both force-set and force-unset, as well as do-nothing
(the default).  Just reading your docs patch, there appears to
be no force-unset, yet that's something you might need where
mod_include invalidates headers set by an app or backend.

Yes, you can force unset by other means.  But if mod_include
is going to offer its own directives, I think they should
include that option.

-- 
Nick Kew

Re: svn commit: r905406 - in /httpd/httpd/trunk: CHANGES docs/manual/mod/mod_include.xml modules/filters/mod_include.c

[Graham Leggett <mi...@sharp.fm>]

On 01 Feb 2010, at 11:24 PM, Nick Kew wrote:

> However, given what you've implemented, I think it would perhaps
> benefit from a third state for ETag and Last-Modified.  Namely,
> allow both force-set and force-unset, as well as do-nothing
> (the default).  Just reading your docs patch, there appears to
> be no force-unset, yet that's something you might need where
> mod_include invalidates headers set by an app or backend.

The patch has two real states, either remove the headers (off, the  
default and current behaviour) or leave the headers alone completely  
and make caching work (on). The off state is the same as current  
behaviour so that it can be backported to v2.2 (where it needs to be  
used for me).

mod_include really shouldn't be trying to set the headers at all, so I  
don't think force-set makes much sense. Last-Modified shouldn't be set  
by the underlying handler, not mod_include, and the same for ETag.  
Unsetting the headers on the other hand does make sense, if the  
content is uncacheable.

For v2.4, simplifying all of this stuff would be a good idea, once  
v2.2 is put to bed.

Regards,
Graham
--