[Grok-dev] Re: goals related to my PyCon Grok talk
faassen at startifact.com
Wed Dec 19 05:27:21 EST 2007
A bit late and before I forget, I'm really happy to hear you are going
to give a talk on Grok at PyCon. I liked the slides I saw of your
previous Grok presentation!
Anyway, my plan for 2008 is to seriously start promoting Grok as a
viable contender in the Python web frameworks world, and beyond. I think
Grok has a lot to offer, mostly thanks to Zope 3 being on top of it.
> As a result of the talk's acceptance, I have the following goals that
> I'd like to tackle, hopefully over the rest of this month and January:
> - I want people to be able to use Grok-the-component-registrar
> without the side effect of slowly importing dozens of Zope web
> modules. This means, if I understand what PvW has been saying,
> that we need to move forward with moving the component-registration
> pieces of Grok out into a separate module, "grokcore", from which
> they can be imported with minimal dependencies (maybe just
> zope.interface and zope.component?). Philip, let me know if I've
> understood this correctly, and if I have, then let me know anything
> I can do to help "grokcore" along!
Actually grokcore is the namespace package that will contain the
reusable parts of grok, such as grokcore.component (the first target).
Philipp, do you have a progress report on this? I think all the major
prerequisites for making this work have been done now (martian,
> - I want Grok and Zope to start much more quickly; a several-second
> startup time really hampers, say, a command-line tool in which you
> just happen to want to use an adapter (as I myself am finding with
> a command-line tool I wrote last week). Getting more speed may be
> impossible, but I'll be profiling just in case to see if anything
> can be done to improve things. (I might look at the recent work in
> speeding up the Plone 3 startup code, to see if any of the issues
> are similar.)
I'd love to see this too. Note that the Repoze people recently blogged
about working on speeding up startup a lot too.
At the Grok sprint in Cologne we had Lennart do a bit of work
investigating selectively turning off ZCML loading to speed up startup
time. Unfortunately that didn't seem to have much of an effect.
More information about the Grok-dev