(side note)
On archlinux that also uses system Python for Sage the
situation is better: importing sage.all in Python2 does work! Even
though the configuration constants are not in the environment! They
are set up properly in sage_env.py that is part of the SageMath
installation. And more precisely $SAGE_LOCAL fallsback to the default
"/usr" which is what it is intend to be on archlinux.
I don't understand why on Debian $SAGE_LOCAL is set to None and
managed by environment variables.
Le 10/01/2019 à 04:47, Nathan Dunfield a écrit :
I am a developer of the Python package "snappy" which acquires extra
features when imported inside of Sage. Some of our Linux users have
recently reported difficulties with SnapPy on machines where Sage was
installed by the standard package manger making use of the system libraries
and in particular the system python. I have reproduced the problem on
Debian Sid but it also manifests itself on Fedora 28 [1]. The root of our
problem is that in this situation:
python -c "import sage.all"
results the error:
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/usr/lib/python2.7/dist-packages/sage/all.py", line 71, in <module>
from sage.env import SAGE_ROOT, SAGE_SRC, SAGE_DOC_SRC, SAGE_LOCAL,
DOT_SAGE, SAGE_ENV
File "/usr/lib/python2.7/dist-packages/sage/env.py", line 123, in <module>
_add_variable_or_fallback('SAGE_ETC', opj('$SAGE_LOCAL', 'etc'))
File "/usr/lib/python2.7/dist-packages/sage/env.py", line 103, in
_add_variable_or_fallback
value = sep.join(components)
TypeError: sequence item 0: expected string, NoneType found
whereas
sage -python -c "import sage.all"
works as expected. Currently, our Python package tests for whether it is
running in SageMath by:
_within_sage = False
try:
import sage.all
_within_sage = True
except ImportError:
pass
One solution for us (I think) would be to simply catch and ignore all types
of exceptions in the above block, but I was wondering if the current
behavior of 'python -c "import sage.all"' was expected or is this possibly
a bug? Or perhaps there is a work-around of some kind?
Thanks,
Nathan
P.S. In the Debian package "sage" does not accept the "-pip" flag, even
though installing the "sagemath" package does pull in "pip" (tested on a
clear Sid Docker container). Why is this? I ask because our current
instructions for installing SnapPy into SageMath use "sage -pip".
[1] See the second comment, not the original report, on
https://bitbucket.org/t3m/snappy/issues/14/
--
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.
