httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Justin Erenkrantz <jus...@erenkrantz.com>
Subject mod_cache backports was Re: mod_dir and mod_cache
Date Wed, 04 Aug 2004 19:29:34 GMT
--On Tuesday, August 3, 2004 10:29 PM -0400 Bill Stoddard <bill@wstoddard.com> 
wrote:

> mod_cache, mod_mem_cache and mod_disk_cache are experimental modules in 2.0,
> so I am going to bypass the votes and just start backporting fixes. Please
> review as they go in. If something breaks, we'll fix it. Mmmm K?

I'll note that without the rewritten apr_file_gets() in APR 1.1 
mod_disk_cache's performance will continue to suck.  (And will likely still 
stink on Windows as that same patch needs to be applied to all other OSes 
too...)  And, you need the ap_internal_redirect (?) patch to cache sub-reqs... 


I really think people are going to have to go to 2.1/2.2 to get the best 
caching performance - it won't be possible to shove it all back into 2.0. 
That's why aiming for a GA in the next few months is a brilliant idea.

I finished some radical rewrites to mod_cache to have it use the provider API. 
It cuts down a lot of code and makes things much cleaner and simpler.  But, it 
removes the caching hooks entirely.  (Expect that patch in a few minutes.) 
Backporting that isn't a viable option though.  It doesn't really help 
performance, but does help readability and extensibility of the caching code.

I've also got some patches to speed up mod_negotiation significantly as well. 
I had to leave last night before I could fully test it and prove that it 
helped.  (A side effect of my test scenario is that it stresses 
mod_negotiation a lot.)  I've also been able to identify bottlenecks in 
mod_setenvif and mod_auth_digest.  So, I have to think about how best to 
resolve those, but I have some ideas that I need to sort out before posting 
more.  -- justin

Mime
View raw message