Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:282
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Reply-To: ilia@prohost.org
To: "lists@holsman.net" <lists@holsman.net>,
 internals@lists.php.net
Date: Tue, 25 Mar 2003 16:55:15 -0500
User-Agent: KMail/1.5
References: <200303251611.12012.ilia@prohost.org> <3E80CD8D.7080505@holsman.net>
In-Reply-To: <3E80CD8D.7080505@holsman.net>
Organization: Prohost.org
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-ID: <200303251655.15492.ilia@prohost.org>
Subject: Re: [PHP-DEV] Preliminary Apache 2 benchmarks
From: ilia@prohost.org ("Ilia A.")

On March 25, 2003 04:43 pm, lists@holsman.net wrote:
> resend due to 30k issue.
>
> > Attached is the patch (ap2.txt) that prevents AP2 handler from constantly
> > flushing the data as soon as it is passed. As you can see in the
> > benchmarks this more the doubles the performance when many writes occur.
> > I am hoping Ian can explain the need for this flush, since without it the
> > code would be clearly faster.
>
> in the general (like 99.9%) of the case you don't need to flush.
> I think disabling it will be a good thing.

We do give the ability to the user to force a flush is they so desire, so I 
think disabling the automatic flushing of data would be a good idea 
especially since performance benefits are so significant.

> the original version I committed actually waited until the entire page was
> parsed before sending anything up the chain.
>
> this is slightly different to what it currently does, and I'll try to see
> if I can see a difference in this approach (compared to sending the buckets
> up the filter chain)
>
> if there is a difference, I'll add a new directive to buffer the response
> up.
>

Ok, let me know what you tests end up showing, I'd definately like to see some 
sort of a solution for this issue appear in the upcomming 4.3.2 release.

Ilia