that is a very valid point, but it seemed that Scott has homogeneous
environment: Debian/Ubuntu so my post was relative to the original request.
I agree that when you throw Windows/MacOS into the mix things
become "interesting". But then it's better when your developers develop on
server/platform they are going to be using, using same stack they going to
face in production etc. It all depends on requirements and current
practices in company.
Well, you certainly want a desktop-orientied Linux for users, so you
chose ubuntu - but then on the server you go with a more stable debian
system. Even though the both have the same technical and even package
management-base, they are still incompatible wrt to package versions for
python.
And other constraints such as Photoshop not being available for Linux
can complicate things further.
- keeping track of recent developments. In the Python webframework
world for example (which the OP seems to be working with), things move
fast. Or extremly slow, regarding releases. Take Django - until 2 month
ago, there hasn't been a stable release for *years*. Virtually everybody
was working with trunk. And given the rather strict packaging policies
of debian and consorts, you'd be cut off of recent developments as well
as of bugfixes.
that definitely becomes tricky however not impossible to track. You do need
a common snapshot for all developers to use anyway - so why not just
package it up?
I do, but based on Python eggs. They are platform independent (at
ultimo, you can use the source distribution, albeit that sux for windows
most of the time), and as I explained in my other post - things are
moving in the right direction.
Don't get me wrong - I love .deb-based systems. But if using them for my
development means that I have to essentially create a full zoo of
various packages *nobody else* uses - I rather stick with what's working
for me.
Diez
--
http://mail.python.org/mailman/listinfo/python-list
