On 17.06.2007, at 22:25, Roger Ineichen wrote:

Hi Juergen

Roger Ineichen

-----Ursprüngliche Nachricht-----
Auftrag von Jürgen Kartnaller
Gesendet: Sonntag, 17. Juni 2007 06:43
An: zope3-dev@zope.org
Betreff: [Zope3-dev] Re: AW: publisher performance

Roger Ineichen wrote:
Hi Jürgen

Betreff: [Zope3-dev] publisher performance

With this new version I also measured the time with zope
as a trunk
checkout (no eggs involved).
The publisher is now twice as fast as it was before !

I'm writing this just to show everyone what can happen if
not enough
care is taken in really critical parts inside the zope core.
newInteraction is called exactly once for each request but
was taking
50% of the time (without eggs) for the publisher.

What do you mean with and without eggs? Do you mean the flat dotted
page name structure used in eggs? Does the egg package structure
perform different in some way? Or do you mean something else?

With "without eggs" I mean I used a trunk checkout for zope.

With eggs I mean I use the eggified packages from zope.

Yes, I understand this, but I'm curios about you commit message:

checkin 76706 says:
Removed stack extraction in newInteraction. When using eggs this is an
extremly expensive function. The publisher is now more than 10 times faster
when using eggs and about twice as fast with a zope trunk checkout.

Why makes this improvment eggs distribution 10 time faster and
the trunk only 2 times?

Do I get this right? Do we pay the flat dotted package structure,
which eggs bring with, pay with slower excecution time?

hi roger

when it comes to module file introspection like the linenumber extraction in the getStack call, the egg version seems to be slow, because there are a lot more directories on the pythonpath, additionally i can imagine, that the effort to find the file for a module is much higher when having a lot of namespace packages

so yes, when tracebacks are generated, the egg version is slower, there might be other places where eggs are slower too, but we didn't find any up till now

Roger Ineichen

Zope3-dev mailing list

Zope3-dev mailing list
Unsub: http://mail.zope.org/mailman/options/zope3-dev/bernd.dorn% 40lovelysystems.com

Zope3-dev mailing list
Unsub: http://mail.zope.org/mailman/options/zope3-dev/archive%40mail-archive.com

Reply via email to