[Pkg-exppsy-maintainers] Fwd: PyEPL Release?

[Per B. Sederberg]

Hi folks:

It looks like we're moving towards a maintenance release of PyEPL.
The next release after that will hopefully be a complete
refactorization that will provide windows support.

Best,
Per


---------- Forwarded message ----------
From: Per B. Sederberg <persed at princeton.edu>
Date: Wed, Jul 2, 2008 at 8:29 PM
Subject: Re: PyEPL Release?
To: Alec Solway <asolway at seas.upenn.edu>
Cc: Josh Jacobs <jojacobs at mail.med.upenn.edu>, Jeremy Manning
<manning3 at mail.med.upenn.edu>, "Christoph T. Weidemann"
<ctw at cogsci.info>, Michael Kahana <kahana at psych.upenn.edu>


Hi Alec:

I've been wanting to talk to you about a release, but haven't had time
with this reply paper I'm writing with Mike.

The main motivation for another release, in addition to the one that
you mentioned, is that there has not been a release since Leopard was
released and we don't have a fully-tested OSX 10.5 install.  I'm
guessing the old one made for 10.4 works to some degree or else this
would have come up earlier.

The issue on my end (well, one of the issues, at least), is that I
don't have a Mac here that has Leopard, so you'll have to be fully in
charge of making the binary installer of Leopard (and you might as
well also be in charge of making it for Tiger because I don't have
good access to any OSX machine anymore.)  Aaron was supposed to have
left instructions for how to do this.

To answer your other questions.  Yes, pyepl is on a public server and
uses the git distributed version control system:

http://git.debian.org/?p=pkg-exppsy/pyepl.git

also look here:

http://alioth.debian.org/projects/pkg-exppsy/

for the experimental psychology project.

You can download and work on it all you want without any sort of
sign-up, but to commit you'll have to become a member of the
experimental psychology developers group on alioth.  Christoph is
familiar with this, as well as the basics of our development practices
(working in branches, rules for commit comments, etc...), which will
eventually be part of a development document.

There have not been tons of changes beyond the last CVS state other
than the addition I made to play movies.  There's some new sound code
that I haven't merged into the master branch, too, but that's not
fully tested.  It should, however, be part of the release because it
has a handful of improvements and new features.

As for testing, the whole of the first version of the pyepl was
written without unit tests, which makes it really hard to add them in.
 The basic approach I think we should take is that all new code should
have unit tests.

Given that we have no full test suite, the best thing to do before
posting a new release is to test it with as many experiments as
possible on both OSX and linux.

Finally, I've started a new branch with a complete refactorization of
the pyepl code base.  That is far from being finished, but the goal is
to make the code more modular and make it easier to provide Windows
support.  Both of those are for the 1.1.0 milestone.

To summarize and finalize:

1) I'm in full support of a release.
2) I'm too busy right now to do the tiny bit of cleanup before it can
happen, though I can do it once my reply paper is submitted (i.e.,
within a week or so.)
3) I will supply you with a tarball of the source from which you can
make the binary installer for the OSX and the debian folks can make a
binary installer for Debian/Ubuntu.
4) Everyone will rejoice because it will be the first release in over a year.


Talk to y'all soon,
Per


On Wed, Jul 2, 2008 at 4:20 PM, Alec Solway <asolway at seas.upenn.edu> wrote:
> Hey Per,
>
> I know you've been maintaining the PyEPL codebase (right?) and I was
> wondering what it would take for a new release. It seems that some of the
> experiments we have posted on our own website rely on a newer version of the
> code than the latest release and people (well, just one lab so far) are
> beginning to notice. Is the code living on a public server now? Has there
> been a lot of changes from what's currently living on our CVS? Is any
> version stable? And how would we know? i.e. I know you said there were
> no/limited unit tests, what's been the previous policy of pushing new
> releases out?
>
> Any insight you can provide would be of great help!
>
> Best,
> Alec
>