[Nix-dev] Re: Python 3

[Marc Weber]

On Wed, Feb 18, 2009 at 01:02:58PM +0100, Peter Simons wrote:
> Hi Marc,
> 
>  > What is Python 3 ?
> 
> it's the latest release of Python -- version 3.0.1, to be accurate.
> 
>  > I "stalled" cause everything worked I had need for.
> 
> Well, but as it is, we have two different expressions for Python 2.5 in
> Nix. That state is not a particularly satisfactory.

My view on this:
I'm using composableDerivation. I've posted it about enough. You can
find doc refs within the composableDerivation attrs in all-packages.nix.
Some people don't like it because you do pass and inherit settings in
some hidden way. We have two expressions cause
a) the existing didn't work for me. I wanted to install some
   applications and I didn't suceed so I started from scratch.

b) Due to the "foreign" composableDerivation style I didn't remove the
  old code cause someone might have trouble reading it.

Using "my" expressions you can install in two different ways:
* use the wrapper (defining NIX_PYTHON_SITES for you) or
* install python and the libraries on its own settings either the new
  NIX_PYTHON_SITES var or the already existing PYTHONUSERBASE env to
  tell python where to find the modules.
I've ony tested the first one. The first one is the the one you'll have
to use when passing python as build inputs to other software unless you
want to create temporary user profiles only to support python and some
import statements..

If you fix the default expressions thats fine with me. Than we can
remove my code. If you tell others they should switch using my
expressions thats fine as well.

Right now I'm trying to switch back ( from upstart-0.5) to upstart-0.3
on my virtual server causing still some trouble. That's why I've
installed pexpect and mechanice to automate debugging tasks. Keeping
many branches in sync is just too much effort.

Apart from that I'd rather work on a Wiki so that we can start
documenting states of
  pyhton (2 versions)
  ghc (...)
  composableDerivation
  applyAnd* funcions etc.

I don't think we're ready yet to say the one way is the one true one for
all use cases. But putting a comparison online with some explanations
etc would be an advantage.

So I want to say I won't put any more efforts into these expressions at
the moment as long they work.

If you have some questions you can ask anytime (by mail or on irc)
whatever way you prefer.

Any suggestions to improve the unsatisfacotry situation are welcome.
I'll try to support you. My time is limited though at the moment though.

Sincerly
Marc Weber