Sage on OSX 10.10

[Volker Braun]

I have a working Sage on OSX 10.10. I suggest to release that shortly, in case anybody else made the mistake of upgrading soon after the initial Yosemite release ;-)  Please review

http://trac.sagemath.org/query?status=needs_review&keywords=~yosemite

and any outstanding blocker bugs...

[John H Palmieri]

When I use all of these fixes on a somewhat old OS X 10.10 machine and run 'make ptestlong', all tests pass. But if I set SAGE_CHECK=yes, then I get many failures in test suites: flint, git, gsl, m4rie, pari, and iml. I think that iml always fails its tests on OS X (in my experience), but others look new. Log files: http://sage.math.washington.edu/home/palmieri/misc/OSX10.10/

--
John

[Volker Braun]

See also what I wrote in http://trac.sagemath.org/ticket/17204#comment:7

I had already reported the flint one a while ago. Git works for me. I have already made tickets for gsl and m4rie.

[Francois Bissey]

In the case of iml, the upgrade to 1.0.4 should solve the failing test problem.
In 1.0.3 and under there was always something that was bound to fail on OS X.
At least that’s how I remember it.

Not sure when I will give the whole thing a spin. I do not see a window of opportunity
before Wednesday (Tuesday for most of you).

François

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

[Volker Braun]

bump

[Volker Braun]

This is still blocking the next beta release..

http://trac.sagemath.org/query?keywords=~yosemite&order=priority

On Friday, October 24, 2014 2:39:45 PM UTC+1, Volker Braun wrote:

[Samuel Lelievre]

Hi,

Under OS X 10.10 Yosemite, with then without homebrew's gcc 4.9.1,
I tried and failed to build Sage.



1. With homebrew's gcc 4.9.1 installed.

Starting from Sage 6.4.beta6 I merged
#17176  u/vbraun/gdb_on_yosemite
#17169  u/vbraun/upgrade_to_gcc_4_9_1
#17204  u/vbraun/osx_yosemite_libtool_version_detection

The compilation ends with some errors, including the
following lines towards the end:

----------
>>> Trying to download http://www.sagemath.org/packages/upstream/setuptools/setuptools-3.6.tar.gz
Traceback (most recent call last):
  File "<stdin>", line 12, in <module>
  File "/Users/s/builds/sage/local/lib/python/urllib.py", line 1399, in <module>
    from _scproxy import _get_proxy_settings, _get_proxies
ImportError: No module named _scproxy
----------

This seems to indicate that building Python succeeded, but urllib
is not working because of a missing _scproxy. Indeed, when I run
Sage's python:

----------
$ ./sage -python
Python 2.7.8 (default, Oct 28 2014, 22:07:12)
[GCC 4.9.1] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import urllib
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/s/builds/sage/local/lib/python/urllib.py", line 1399, in <module>
    from _scproxy import _get_proxy_settings, _get_proxies
ImportError: No module named _scproxy
----------

2. I uninstalled homebrew's gcc, ran 'make distclean', then 'make' again.

This failed to build gcc, the relevant logs are here:

http://carva.org/samuel.lelievre/t/20141029_osx1010_sage64b6_install.log.tgz
http://carva.org/samuel.lelievre/t/20141029_osx1010_sage64b6_pkgs_config.log.tgz
http://carva.org/samuel.lelievre/t/20141029_osx1010_sage64b6_pkgs_gcc-4.9.1.log.tgz
http://carva.org/samuel.lelievre/t/20141029_osx1010_sage64b6_gcc491_config.log.tgz





I then uninstalled homebrew's gcc, ran 'make distclean', then 'make' again.

This failed to build gcc, the relevant logs are here:

http://carva.org/samuel.lelievre/t/20141029_osx1010_sage64b6_install.log.tgz
http://carva.org/samuel.lelievre/t/20141029_osx1010_sage64b6_pkgs_config.log.tgz
http://carva.org/samuel.lelievre/t/20141029_osx1010_sage64b6_pkgs_gcc-4.9.1.log.tgz
http://carva.org/samuel.lelievre/t/20141029_osx1010_sage64b6_gcc491_config.log.tgz

[Samuel Lelievre]

Sorry, the last 6 lines in my last post are there twice,
please ignore the repetition.

[kcrisman]

Maybe the next beta should not be blocked by this?  It's only final releases that have blockers, I guess... unless you mean it's because it doesn't work on your laptop, but I assume (perhaps incorrectly) that sage.math is still the "official" release machine.

See also http://wiki.sagemath.org/SupportedPlatforms which I minimally updated with respect to this just now, but which probably needs some more significant updating as I think we may not have all the Roman-named machines anymore.

- kcrisman

[Volker Braun]

On Wednesday, October 29, 2014 12:45:15 PM UTC, kcrisman wrote:

See also http://wiki.sagemath.org/SupportedPlatforms which I minimally updated with respect to this just now, but which probably needs some more significant updating as I think we may not have all the Roman-named machines anymore.

If you want me to take off OSX from the list of supported platforms then we can do that of course. I don't really mind but I suppose others do ;-) 

[Volker Braun]

Gcc picks up parts of your homebrew install, you must at least rename /usr/local before you can build anything with homebrew installed there.

[Dmitrii Pasechnik]

On 2014-10-29, Volker Braun <vbrau...@gmail.com> wrote:
> ------=_Part_5637_1547295187.1414591060718
> Content-Type: text/plain; charset=UTF-8

Probably OSX 10.10 can wait for the next (sub)release...
>

[Volker Braun]

I can't even test tickets on OSX without the gcc update, because our only buildbot is running on 10.10.

IMHO the only thing that CAN wait is beautification of the scripts or repacking the gcc tarball to save some disk space...

[kcrisman]

I can't even test tickets on OSX without the gcc update, because our only buildbot is running on 10.10.

IMHO the only thing that CAN wait is beautification of the scripts or repacking the gcc tarball to save some disk space...

So "OSX 10.6 (Snow Leopard) 64-bit, with XCode 3. Tested on bsd" is no longer accurate?  Hmm.  I mean, many of us test on other versions of OS X, but it's hard to provide a dedicated buildbot slave. 

[Volker Braun]

On Thursday, October 30, 2014 12:23:57 PM UTC, kcrisman wrote:

So "OSX 10.6 (Snow Leopard) 64-bit, with XCode 3. Tested on bsd" is no longer accurate? 

Yes, no longer accurate since the bsd.math machine died.

[Samuel Lelievre]

I uninstalled homebrew and tried again, the build was successful.
I will now build sage-6.4.rc0. Thank you for making this work!

2014-10-29 15:07:19 UTC+1, Volker Braun: