[Grok-dev] PyCon 2008 Grok Sprint, day 2

Brandon Craig Rhodes brandon at rhodesmill.org
Wed Mar 19 12:11:01 EDT 2008


Our small team seemed to feel more frustration as we pushed through
the second day of our sprint yesterday.  The discovery that the edits
done by the team to the tutorial on Monday had to be re-done yesterday
to the Subversion file itself seems to have been a big blow to morale;
I ought to have known more about the web site before having people go
in to improve our documentation. :-/ But, then, I should have done a
lot more planning before attempting to sprint-coordinate; I had far
too few little tasks for people completely new to the framework who
knew nothing about it.  I hadn't realized that lots of little tasks
are much better than one big task for a sprint.

But we did accomplish a few things on Tuesday: the
"grokcore.component" package is now working, passing all of its tests,
and I can use its Adapters in a non-Grok application successfully.
The changes to the tutorial, as noted above, did get eventually
committed to Subversion.  Brad Allen looked over the Grok
documentation and made the excellent point that our site looks
basically empty of documentation for the first three or four clicks
that a user does down into its hierarchy (!).  And Reed O'Brien and
Robert Marianski made a solid improvement to Grok by removing the
horrible "*Base"-class auto-detection from Martian, which was one of
the bugs on the Grok launchpad.

The day ended with saying goodbye to nearly everyone as they prepared
to leave Chicago, except myself (who today am moving to the sprint
room that has the WSGI/Pylons/Turbogears people so that I'm not in a
big empty room by myself with the Pyglet people), and Brad Allen (who,
to expand his range of experiences, will be visiting some other sprint
groups today).

Today I am therefore more or less flying solo, and have some goals:

 - Learn about how to move my own stuff to "bzr", since it's become
   clear to me over the last few days that it is the One True Revision
   Control System.  I'm glad my instincts told me not to jump to "git"
   after seeing Linus's Google Video talk about it, but to stay with
   Subversion and move when something even better came along.

 - Think about DBSprockets.  The guy who wrote it is at the other
   corner of the sprint group I'm sitting with, and he'd love it to
   work with Grok.  My creating a "megrok.dbsprockets" would make his
   day. :-) And then Grok would have an object browser for admins, and
   yet another way to generate forms.

 - But DBSprockets needs WSGI, and "grokproject" is still based on the
   Zope web server.  Maybe I should look at making "grokproject" work
   like "zopeproject".

 - But first: I will branch Grok itself and make it work with
   "grokcore.component", so that I can feel kind of done about that
   whole effort.

 - I would like to write some docs about using "grokcore.component" in
   a normal app, but I'm not sure which way to tell them to invoke the
   grokking of their app.  In my little sample app that I tried it
   with, I just ran "grokcore.component.grok(...)" on the
   application's top-level package, and the adapters worked (if I used
   grok.context(), but not grok.adapts(), about which see my separate
   email on the list).

 - I need to write a release script for myself.

 - What was I going to do with "megrok.trails", again?  I can't
   remember what I was going to improve before releasing it. :-)

-- 
Brandon Craig Rhodes   brandon at rhodesmill.org   http://rhodesmill.org/brandon


More information about the Grok-dev mailing list