New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
numpy-1.7.0_pre20121217 fails with sage-5.5_rc0-r1 on ~amd64-linux #179
Comments
Updating to the new eclass is the best idea I think. I thought the stub I used from 1.6.2 in the tree was already migrated. Obviously I am wrong. Will work on that shortly. |
I see I opened a can of worms here. That was not the intention. But I suppose eventually it would have had to be done. Everything I've tried now seems to be working. |
Yeah that's a can of worm. I think I will still work on both eclass but will do something about python-3.3 with the old eclass. 1.7.0rc1 will have to be tested against python3.3 there have plenty of python 3.3 fix in rc1 so it may be supported now. |
https://bugs.gentoo.org/show_bug.cgi?id=450910 leads me to think that the problem may actually lie with python.eclass. Things are properly handled at the ebuild level so it must be the eclass. As said earlier 1.7.0_rc1 may just work with 3.3. |
I see. In the numpy ebuild I removed the restriction from building 1.7.0_rc1 against python-3.3 and there seems to be no problem now. I have both 2.7 and 3.3 compiled versions of numpy. |
Ok, I'll remove the restriction in rc1. The next interesting bit is whether or not scipy will compile against 3.3.... |
Here on ~amd64-linux, scipy-0.11.0 does build against python-3.3 |
I am a bit reluctant to import scipy in the overlay but if we get complaints about building when python 3.3 is installed I will provide a way out. |
For those who don't follow http://trac.sagemath.org/sage_trac/ticket/11334 the fun never ends. inline_fortran doctest is broken until the new doctesting framework from http://trac.sagemath.org/sage_trac/ticket/12415 lands. This is huge.... |
For all purposes I think this is fixed. Someone will have to port numpy/scipy to the new python eclasses but that's another story. |
On ~amd64-linux, python-3.3.0 is unmasked. In building the subject numpy I get
The usual tricks to get numpy to build against just python-2.7 don't seem to work. At least I haven't been successful. So is numpy now going to have to be made compatible with the distutils-r1.eclass for things to work? Some quick tests here in adding the eclass to numpy, with a few changes, did allow it to build. Or perhaps masking python-3.3.0 is the better way to proceed.
The text was updated successfully, but these errors were encountered: