[Yt-dev] binaries

[Matthew Turk]

> This isn't a bad idea.  I just read through the description of virtualenv.
> Can you describe shortly how we would use this?  If I had a yt installation
> on some machine in /home/britton/local of something, and someone just made a
> copy of that, would they need virtualenv to make it run from their own
> directory or could it go on it's own?  or is this not at all how this would
> work?  Sorry for the stupid question.

It's not a stupid question at all.  What -- I believe -- virtualenv
would provide to us is a means of getting around hardcoding of paths.
If you look in any of the scripts yt installs, the path is hardcoded
at the top.  VirtualEnv would simplify the process of getting around
that.  (VirtualEnv is relatively light weight, but pretty sweet in
that it lets you move between different complete installs of python.)

So if we provided binaries as of right now -- in the form of a big
tree of files -- they would not work because of this hardcoding.  But
virtualenv would help out with that, in that it knows how to
transplant.

> On another note, I read that blog post you referred to at the beginning.  I
> can see where you're coming from wanting to speed up the gratification for
> the new user.  In the context of that post, I would say you (we) are already
> over that initial hump.  That said, this certainly won't hurt.

I think we are getting there...  I think we should aim for a 1.5 (or
is it 2.0 by now?) release by May 15.

http://yt.enzotools.org/report/12

this, and documentation of new features and a feature list are all we need.

-Matt