-
Committed 10.6-specific python2[56] packages incorporating apple's patches. Thanks for tracking down the problem/solution!
2009-10-31 19:53:56 UTC in Fink
-
Managed to apply it. However, monipol reports that the bug is apparently not present on 10.5 either I wonder if this is a 10.6-specific mess, and if this patch is safe for 10.[45] users (or else will require a separate python package for 10.6 users).
2009-10-22 16:24:11 UTC in Fink
-
(Forgot to mention, my failed attempt to reproduce the bug was on 10.4/ppc)
2009-10-22 16:15:54 UTC in Fink
-
I was unable to apply their patch to our current python2.6 successfully. I got a "?" response from ed and there was no change made to the sources.
However, when I tried their test-case (import os; os.getgroups(); according to http://trac.macports.org/ticket/21559) fink's current python25-1:2.5.4-1 and python26-1:2.6.2-2 give sane results rather than error messages.
2009-10-22 16:13:16 UTC in Fink
-
Yes, I had worked up a ghostscript package that also enabled the libgs library (other packages are starting to look for it as ghostscript back-end rather than just using the ghostscript executable). Also, I had noticed some problems with getting the internal libraries linked correctly (they were getting mixed up with the same-named libs supplied by fink) so I needed some hackery to solve it. See:
2009-08-13 17:52:02 UTC in Fink
-
In addition, notice that the existing package has an option about whether to use atlas or not. With yours, you are forcing atlas (which apparently some users do not want or cannot use). More confusingly, you are forcing it into the package-name that previously did *not* have it, so (again as my previous comment) get unexpected changes just based on updating their existing package.
2009-07-08 09:46:18 UTC in Fink
-
I can tell just by looking at the .info that it won't work on case-sensitive filesystems. "MacOS" and "MacOs" are not the same.
Why waste so much time retyping the same perl one-liner for each of the files to patch instead of doing all the files together in a single command?
I'm tempted to reject this package out-of-hand because it drastically changes the behavior compared to the previous...
2009-07-08 09:34:41 UTC in Fink
