I've been obsessing lately over how quickly I can develop and improve web
applications.  Being on soft money for years will do that to you, I
suppose.  Anyway, being a devoted python fan, a shocking wave of new
toolkits has arrived recently: django, turbogears, and subway (oh, and
zope3 has been out for a while, which defintely deserves a look).  I
haven't used ruby on rails or catalyst but I'm told they all have a
similar imprint:  full and integrated support stack for webapps from
OR-mapping to controllers to templating, resulting in much faster
implementation cycles.

There's something about turbogears that I'm finding particularly
compelling, but any of these looks potentially better than what I've been
doing, which includes quixote and PTL for webapp control and templating,
respectively.  I still love quixote+PTL, and it's likely my code-design
skills are wanting anyway, but I'm finding now that my two main webapps
are both more than 18 months old, that each comprises many creaky corners
and repetitive tasks that I suspect might wash away in one of these newer
toolkits ("megaframeworks"?).

Maybe more importantly, it seems like these new tools are being optimized
for web2.0-style data publishing concurrent with and sometimes seamlessly
alongside for-human interfaces.  Though that kind of thing is fairly
easily backported to 2002-03-era web frameworks it seems radically simpler
in the 2004-05 ones.

Has anybody out there jumped on one of these new toolkits?  If so, am I
just being distracted by the shinyness of the new stuff?  Or are you
seeing the kind of development-time speedups these things (and the fawning
hype surrounding them) promise?  Also, as I don't pay much attention to
the .NET stack and tools like coldfusion, is it possible that my Free
Software blinders have me believing these "new developments" are just
catching up to where proprietary toolkits have been for years?  Seems
possible.

And, heck, what's coming in 2006? :)

   -Dan