I think we need to reign in the personal attacks in this thread.  Labels
like "'so-called' professionals" and "incompetent" do nothing to further the
discussion.
Now that Qiang Xue has re-ran the tests against RHEL 5 using Apache 2 and
the latest PHP release, it's evident that Yii Framework is quite a bit
faster than Zend Framework.  I took a look at the source code and there are
a few things that Yii does that ZF does not:

- Autoload by default.  Because this represents a BC change, this is planned
for ZF for 2.0, IIRC.
- Multiple classes in a single file.  I doubt ZF will ever do this, as it is
a maintenance headache.
- Dump a bunch of the most commonly-used classes into an easily cacheable
file with comments stripped out.  At a minimum, I think it's worth
considering having a Zend_Tool-based generator that accepts a list of
components (and has a sane default) and spits out a single file (perhaps
named Zend.php, since it's auto-generated).  Then users would just need to
set up their include path and require_once 'Zend.php';

-Matt

On Mon, Nov 3, 2008 at 9:13 AM, ekerazha <[EMAIL PROTECTED]> wrote:

>
>
> Isaak Malik-3 wrote:
> >
> > Also, I've run only 2 benchmarks per framework and they were all warm
> > benchmarks, in theory real benchmarks should be performed several times
> > and
> > the average should be calculated out of those. And if we do that for my
> > benchmarks this brings up the following results:
> >
> > Yiilite without APC:
> > 141.08
> >
> > Yiilite with APC:
> > 153.69
> >
> > So using APC still performs better.
> >
>
> Well, being fussy :-) you have:
>
> ---
> Yiilite:
> 153.15 RPC without APC
> 129.01 RPC without APC
> ---
>
> You can't calculate a credible average value beetween two values that are
> so
> different, you have a 20% Delta on the same app, that's just not
> credible...
> obviously something went wrong with your tests.
> --
> View this message in context:
> http://www.nabble.com/Framework-speed-shotout----question-tp19914787p20306654.html
> Sent from the Zend Framework mailing list archive at Nabble.com.
>
>

Reply via email to