[PEAK] Packaging peak apps

Phillip J. Eby pje at telecommunity.com
Thu Sep 16 17:05:49 EDT 2004


At 03:45 PM 9/16/04 -0500, Stephen Haberman wrote:
>Took me a little while, but I finally got to splitting those that reload bug
>up.
>
> > * File a bug report for the PEP 302 issue (loader not carried through by
> > reload function), and assign it to me
>
>https://sourceforge.net/tracker/index.php?func=detail&aid=1029475&group_id=5
>470&atid=105470
>
> > * Attach the patches for import.c and test_importhooks.py; leave off
> > everything about zip importing.
>
>Done.

I just sent you a reply via SF about this, but it's more convenient for me 
to discuss it here and post a summary there, so if you prefer, let's 
continue the discussion here.


> > * Verify that this reduced patch set is correct; i.e. its test case should
> > fail under unpatched 2.3 and 2.4, and succeed under patched 2.3 or 2.4.
>
>Yep.
>
> > * I will apply the patch for 2.4 and backport to the 2.3 maintenance
> > branch, so that this can go in 2.3.5 as well as 2.4
>
>Sounds good. I had to update my patch slightly, mainly for this new "newm"
>variable, so you might have to backport that as well.

Yeah, I'm having trouble with your 'path' change, too.  But the other issue 
has nothing to do with the C variable named 'newm'.  It's strictly to do 
with the fact that PEP 302 loaders aren't required to reuse the existing 
module for reload.  This is utterly broken, and I've posted to python-dev 
about it, but gotten no response.  I intend to repost there soon.

Together, maybe we can get the base issue fixed before 2.4 beta, which 
would allow people to package PEAK apps using py2exe and Python 2.4 (or 
2.3.5, if there is one).




More information about the PEAK mailing list