You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Justin Erenkrantz <je...@ebuilt.com> on 2001/11/16 17:31:37 UTC
Re: can somebody remind me why the threaded.c MPM is still around?
On Fri, Nov 16, 2001 at 08:11:10AM -0800, Aaron Bannert wrote:
> On Fri, Nov 16, 2001 at 09:07:33AM -0500, Jeff Trawick wrote:
> > (the subject says it all)
>
> We just need to decide if the worker MPM is to become the primary
> hybrid threaded/process MPM, no?
+1 for removing threaded MPM.
+1 for activating worker MPM as the default when we have threads.
+1 for renaming all of the MPMs to actually make some damn sense.
> [Doesn't this discussion belong on dev@httpd?]
Yes. -- justin
Re: can somebody remind me why the threaded.c MPM is still around?
Posted by Jeff Trawick <tr...@attglobal.net>.
I just put some votes on a) remove threaded and b) make worker the
default in the STATUS file. Sorry, I left out the "renaming all of
the MPMs" thing... One of you folks can do that if you choose.
--
Jeff Trawick | trawick@attglobal.net | PGP public key at web site:
http://www.geocities.com/SiliconValley/Park/9289/
Born in Roswell... married an alien...
Re: can somebody remind me why the threaded.c MPM is still around?
Posted by Aaron Bannert <aa...@clove.org>.
On Fri, Nov 16, 2001 at 08:43:00AM -0800, Justin Erenkrantz wrote:
> The only additional requirement with worker is condition variables.
I believe the overriding constraints are the fork semantics, but yes
condition variables are another requirement. However, the worker MPM
only uses the APR calls for all locks/threads/condition variables,
so if it's supported in APR it'll work everywhere.
OTOH, I'm -0 for replacing prefork with worker on any platform until
we discuss the tradeoffs a little more. I've read no statistical
distributions of the ways apache is applied, which is what I'd need
to make an informed decision on what MPM with what parameters would
be optimal. I still seem to think that prefork will be better in
the general case, but I could be convinced otherwise.
-aaron
Re: can somebody remind me why the threaded.c MPM is still around?
Posted by Justin Erenkrantz <je...@ebuilt.com>.
On Fri, Nov 16, 2001 at 08:38:22AM -0800, Ian Holsman wrote:
> is there a platform that worker doesn't work and threads does?
Doubt it.
The only additional requirement with worker is condition variables.
-- justin
Re: can somebody remind me why the threaded.c MPM is still around?
Posted by Ian Holsman <ia...@cnet.com>.
On Fri, 2001-11-16 at 08:31, Justin Erenkrantz wrote:
> On Fri, Nov 16, 2001 at 08:11:10AM -0800, Aaron Bannert wrote:
> > On Fri, Nov 16, 2001 at 09:07:33AM -0500, Jeff Trawick wrote:
> > > (the subject says it all)
> >
> > We just need to decide if the worker MPM is to become the primary
> > hybrid threaded/process MPM, no?
>
> +1 for removing threaded MPM.
> +1 for activating worker MPM as the default when we have threads.
> +1 for renaming all of the MPMs to actually make some damn sense.
>
+1 x 3
is there a platform that worker doesn't work and threads does?
> > [Doesn't this discussion belong on dev@httpd?]
>
> Yes. -- justin
--
Ian Holsman IanH@cnet.com
Performance Measurement & Analysis
CNET Networks - (415) 344-2608