Fix the channel used for SymEngine.

Fixed to pull SymEngine + Python bindings from conda-forge.

Also remove SUMPY_NO_CACHE=1 in the Conda build.

I guess SUMPY_NO_CACHE was set in the gitlab CI environment in order to test SymEngine. But this may be the cause of the incredibly long build times....

added 1 commit

• 1ce2dcb4 - Don't allow Conda failure.

Compare with previous version

This makes the Conda build significantly faster. It is still slow compared with the regular Python build though.

changed title from Fix SymEngine channel. to Fix the channel used for SymEngine.

added 1 commit

• 7767ef68 - Remove symengine channel now that everything is in conda-forge.

Compare with previous version

changed the description

Hmm. It looks like the failure is due to the fact that my_syntactic_subs isn't working anymore. I think the version of SymEngine in conda-forge won't let me construct a Subs node.

added 1 commit

• c607de07 - Mark test_matrix_build as skipped under SymEngine.

Compare with previous version

mentioned in commit d4ff9da0

merged

LGTM. An hour for the conda test is much more manageable. :) Thanks for working on this.

mentioned in issue #60 (closed)