I'd actually suggest breaking out other parts that are updated more
frequently, or could be used independently of TurboGears. Like Catwalk,
which was actually separate to begin with. I can understand the desire
to package it with TG for a full-blown system, but I don't see any
other benefit that was gained by tying a separate application into a
framework.

With Pylons, we've already gone for this approach, which lets us have a
very stable core while extending it through WSGI middleware and
libraries. Both of which don't really rely on Pylons even, so they're
re-usable in other setups.

There's a big caveat though that we've run into. First, I'd advise
against using a setuptools namespace package. It's caused more than a
few headaches in installation and dependencies with Paste, so I'd
consider it a last resort type of thing. Second, the more dependencies
you have, the greater the risk of a failed install due to either
setuptools not being able to find the package, or the package
disappearing (guess that's the same).

In our case, we went with a backup option of having every single egg we
use on a download page as a last resort should some package somewhere
in the dependency chain not get properly installed. Alternatively, when
relying on such a conglomerate of packages, the risk of a single
package somewhere in the chain getting a backwards incompatible patch
applied, or some other feature change you relied on; becomes a very
real risk. The download page alleviates this as well by providing a
source for all the egg's known to result in a working system.

Cheers,
Ben


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"TurboGears Trunk" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at 
http://groups.google.com/group/turbogears-trunk
-~----------~----~----~----~------~----~------~--~---

Reply via email to