[development] CVS HEAD, code freeze, zeitgeist

Jeff Eaton jeff at viapositiva.net
Wed Aug 16 21:06:14 UTC 2006

Thanks for the heads up and the opportunity to discuss, Dries.

There's one relatively major patch that I've been involved in -- the
switch from text-based to array-based assembly of node contents for
rendering. I'm very happy with it in general, but after further
examination I think there must be another patch to smooth the rough
edges off one aspect of it (clean and convenient rendering of sub-chunks
of the content array). That's not a show stopper, but I need to get a
patch out for it this weekend as more free time opens up.

The other biggie that is down to the wire is the switch to 'pull' rather
than 'push' workflow for FormAPI. (http://drupal.org/node/77919 for
those who aren't memorizing patch nids). It's a very important change,
one that removes a large number of fundamental roadblocks in the way of
things like programmatic (rather than UI-driven) use of forms to save
and edit data, the clean integration of multi-step wizard-style forms
into core, and so on. There are some rough edges to the patch, and it
does not directly implement any of the awesome 'And I want a pony"
features that it paves the way for. But it is very significant and I
would hate to see it not make it.

The install system is in place, but there are still problems that must
be resolved before it can be used ina a trouble-free fashion by end
users (namely some of the requirements checking stuff, and ways to
support multi-site installations without blowing away the 'clean' copy
of settings.php that's necessary.)

We have core completely patched and ready for jQuery integration (which
is really essential for Drupal's future as a web app platform, IMO), but
we probably won't get it in because of delays in jQuery 1.0's release.

I mention these particular patches/issues/improvements because they all
represent, to me, the current state of HEAD. Full of possibility, but
also rough around the edges. There are tons of new under-the-hood
capabilities, but we're not using them yet. There are lots of new
subsystems, but we haven't yet ironed out all the problems that they'll
face when people try to convert all of contrib over. And so on and so

It's exciting, but a bit nerve wracking. I suppose that these things are
all what gets shaken out after the code freeze, when we can focus on
refining and fixing issues rather than trying to stick new features in. 


More information about the development mailing list