{tools}[GCCcore/8.3.0] hypothesis v4.44.2 w/ Python 3.7.4#9317
Conversation
|
Test report by @branfosj |
|
Test report by @branfosj |
|
Test report by @branfosj |
|
|
||
| builddependencies = [('binutils', '2.32')] | ||
|
|
||
| dependencies = [('Python', '3.7.4')] |
There was a problem hiding this comment.
@branfosj hypothesis still works with Python 2, no?
in that case please use multi_deps as was done for hypothesis-4.23.4-GCCcore-8.2.0.eb?
(see also the email of Kenneth to the mailing list from 1 day ago)
There was a problem hiding this comment.
I've just tried to multi_deps it and I'm hitting an odd error: https://gist.github.com/branfosj/61a8bf5135f2cc422c1f8d545b5da7f3
https://gist.github.com/branfosj/61a8bf5135f2cc422c1f8d545b5da7f3#file-easybuild-hypothesis-4-44-2-20191116-190723-sallh-log-L1124 shows: TypeError: attrs() got an unexpected keyword argument 'eq' which is unexpected. This feature got added to attrs in 19.2 but I'm using 19.3 here, which suggests that we're picking up the attrs 19.1 that is included in the Python 2.7.16 and 3.7.4 installs instead. However, this doesn't happen when I build the Python 3.7.4 specific hypothesis module.
And,
diff -r ../../../../../../EL7/EL7-cascadelake/software/hypothesis/4.44.2-GCCcore-8.3.0-Python-3.7.4/lib/python3.7/site-packages/ ../../../../../../EL7/EL7-cascadelake/software/hypothesis/4.44.2-GCCcore-8.3.0/lib/python3.7/site-packages/
shows that the two lib/python3.7/site-packages directories are identical.
I'm running at version 4.0.1 for framework, easyblocks, and easyconfigs.
There was a problem hiding this comment.
You can't easily override a Python package from the Python installation when using multi_deps, we've run into that earlier this week for something else, basically because for multi_deps to work we don't use $PYTHONPATH but site.addsitedir, and apparently the latter only append, there is no way to prepend (see also https://bugs.python.org/issue7744).
So I guess we'll have to go forward with Python 3-only here.
Is this a dependency for something else?
There was a problem hiding this comment.
I've started work on PyTorch 1.3.1 and this is for that. The pull requests I have done today are all in support of that.
There was a problem hiding this comment.
We're stuck to a single Python version there anyway, due to SciPy-bundle, so OK.
|
Test report by @boegel |
|
Test report by @boegel |
|
Test report by @boegel |
|
@smoors OK to merge? |
|
@branfosj Does this version of hypothesis depend on attrs >= 19.2 then? |
|
Yes. For PyTorch I believe that I do not need such a recent |
|
Ok, then let's go with this one .... |
|
Going in, thanks @branfosj! |
1 similar comment
|
Going in, thanks @branfosj! |
can't do multi-deps here due to needing an updated "attrs" version.,
|
Going in, thanks @branfosj! |
(created using
eb --new-pr)