Sage update

[Montgomery-Smith, Stephen]

On 02/05/2014 11:04 AM, Montgomery-Smith, Stephen wrote:
> On 02/05/2014 12:29 AM, Daniel Smith wrote:
>> I just tested it on my box. It gets a lot further, but crashes when
>> building scipy-0.12.0.p1. The only message in the log is a line
>> indicating that the spkg-install script failed on a line reading 
>>
>> python setup.py setup
>>
>> I'm currently working on digging through that file to see if I can
>> figure out more, but I won't be able to do much before tomorrow, I'm
>> afraid.
>>
>> I've attached the log file for details.
>>
> 
> I and one other person got the same error.
> 
> I did try the patch
> /usr/ports/devel/scons/files/patch-engine-SCons-compat-_scons_subprocess.py,
> but it didn't help at all.
> 
> If you type make again, I find that sage tries skipping the packages it
> failed to build in favor of other packages.  In this manner, you can
> find that matplotlib-1.2.1 fails to build in exactly the same way.

I built the science/py-scipy port.  Comparing the build logs, it seems
to me that sage completely builds the scipy subpackage.  And then python
crashes.  So somehow the problem is not with scipy as such.

You could just take out the error handling in the script spkg-install
inside build/pkgs/scipy.  That would be rather dangerous, but it might
provide for a temporary work around the problem.

To be honest, I am finding FreeBSD-10 to be quite a pain.  It is
something of a moving target.  One of my other ports now fails, where it
seems it used to build under FreeBSD-10 quite successfully.

If any of you have good insights, I would appreciate it.  But I may not
be able to spend a lot more time hunting down this bug until next
summer.  And I would prefer to wait until FreeBSD-10.1 is rolled out,
and things are more stable.