Hi,

interesting.
I have the faint hope that the fix at 
http://trac.sagemath.org/sage_trac/ticket/12161
(which I hope to review this weekend --- it requires building twice a
Sage app on OS X 10.6, and then test these on OS X 10.7) will also
cure this. Even though this new report is on OS X 10.6!

Reasoning:
The underlying problem fixed by the patch at #12161 was, that a system
python binary (/usr/bin/python) was called, instead of the python
binary that comes with Sage. According to some web quick searches,
"_PyUnicodeUCS4_AsUTF8String" is encountered when trying to import
Python extensions into Python binaries which are slightly incompatible
with each other (the difference being about the idea of how to
represent Unicode, as UCS2 or UCS4) ...

The #12161 problem was so difficult to track because it hurts only on
systems that *never* had a Sage installed on it, more precisily that
are missing the ".sage" subdirectory under their user's home
directory.

I remember that in the past, we had problems because nobody had tried
to test whether relocatability still works. I'm not quite sure how
many regular build/regression tests we currently run (except for mere
building and doctesting), and if some relocatability test is among
these (this just can't be "easily doctested" for obvious reasons).
Maybe it would be a good idea to also test Sage rc releases in
environments, where ".sage" has been removed (for each major platform)
--- there may be many other issues, with matplotlib, gap, and any of
the other components that write (or only look) under ".sage". (Testing
this also is pretty hard to do by the doctesing framework, I presume).


Cheers,
Georg

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Reply via email to