In gnome's overlay, we (the gnome herd) are encountering a problem with some packages. Those packages try to use some of the new features available in upstream python-config which are not yet available in gentoo. For now, we are patching out the relevant lines but it is to be expected to hit this problem more often in the future. I've been testing the python-config actually used in ubuntu (from the 2.5 dev package) with python-2.4 and it works well as far as my tests goes. Please advise. Reproducible: Always Steps to Reproduce:
*** Bug 178096 has been marked as a duplicate of this bug. ***
Created attachment 127313 [details] Replacement python-config I'd like to bump this. Most of the gnome 2.20 python bindings will need this. Here's the python-config from ubuntu (which is very similar to the one in python-2.5). It works for me. I'd like to get this into a python 2.4 bump, unless we're sure that python 2.5 will be unmasked very soon. If it's preferable, I can make it return the same info as our current one does if called with no options.
*** Bug 188263 has been marked as a duplicate of this bug. ***
Python herd if you don't move your ass until the end of the week i'll do myself the changes to get this bug done. Thanks
*** Bug 184553 has been marked as a duplicate of this bug. ***
I spoke with hawking about this a couple days ago. Meanwhile python2.5 has come out of p.mask, so for now ~arch is good. However I stressed then and stress again, that we should have a fixed python2.4 as well. Keep reporting ldflags with no arguments but have the include flags and other things possible to be reported with appropriate arguments being passed to it. A starting point is provided in comment #2, with an offer to make the work to report ldflags with no arguments in your help. Why I stress is because we don't want to depend on python2.5 getting stable on time for GNOME-2.20, delaying the stabling of 2.20 again - we really want to get this stable in reasonable time this time around, and having such big other things block it is not good - you don't know what unexpected roadblocks can turn out for python2.5 stabling, so lets please have a 2.4 revbump with the python-config fix that can be stabled rather soon.
Created attachment 129051 [details, diff] compatibility.diff
(In reply to comment #6) > I spoke with hawking about this a couple days ago. Meanwhile python2.5 has come > out of p.mask, so for now ~arch is good. > However I stressed then and stress again, that we should have a fixed python2.4 > as well. Keep reporting ldflags with no arguments but have the include flags > and other things possible to be reported with appropriate arguments being > passed to it. A starting point is provided in comment #2, with an offer to make > the work to report ldflags with no arguments in your help. I agree, we need a fixed python-2.4. I wrote a small patch to make the attached python-config compatible with our old custom python-config. I plan to revbump python-2.4 after speaking with other python guys ;)
Thanks for giving some attention to this bug. I am strongly in favor of a revdump of python-2.4 with a fix too. Can someone explain why the current python-config provided by 2.5 isn't good for 2.4 ? As far as I my tests goes there were no problem.
(In reply to comment #9) > Thanks for giving some attention to this bug. I am strongly in favor of a > revdump of python-2.4 with a fix too. Can someone explain why the current > python-config provided by 2.5 isn't good for 2.4 ? As far as I my tests goes > there were no problem. > It's good for 2.4 - the first attachment is actually python-config-2.5 ;) - , it'll just cause problems for packages that rely on the specific behaviour of our old custom python-config. There aren't many afaik - one example is dev-tcltk/tclpython - but considering my patch is harmless for packages using the python-config, I think it's wise to apply it so that we can fix stuff slowly without breaking anything and providing backwards compatibility.
python-2.4.4-r5 is in CVS with the new python-config. Thanks everyone :)
that's fine with me, thanks for the fix.
Woo hoo! Thanks.
*** Bug 192898 has been marked as a duplicate of this bug. ***
Reopening for stabilization, please stabilize 2.4.4-r5
CC'ing arch-teams. Thanks in advance!
Amd64 out.
x86 done
alpha/ia64 stable
stable on ppc
ppc64 stable
Sparc stable (all tests pass, portage works).
Stable for HPPA.
mips stable.
hum looks like I forgot to add m68k here arm, sh, s390, ping