bump python requirement to <3.8

broadinstitute / longbow

Annotation and segmentation of MAS-seq data

https://broadinstitute.github.io/longbow/

BSD 3-Clause "New" or "Revised" License

20 stars 4 forks source link

bump python requirement to <3.8 #166

Closed jamestwebber closed 2 years ago

jamestwebber commented 2 years ago

python 3.8 is listed in setup.py, so it should be supported. It looks like there are a few test failures as well as syntax warnings:

[x] is not "" is now a SyntaxWarning because literals are not guaranteed a constant identity, needs to be changed to != ""
[ ] tests are failing due to some change to the language, need to investigate. edit this is a pysam issue

jamestwebber commented 2 years ago

ah, it looks like this due to an open issue in pysam, which doesn't support multiprocessing for python>=3.8 for some reason.

In that case it probably makes more sense to remove the listing in setup.py. I know that the tests all pass with 3.7.12 at least, so the requirement can be changed to <3.8

jamestwebber commented 2 years ago

After some investigation (see pysam-developers/pysam#924) it looks like the issue is just due to passing a BAM header object. It should be simple to tweak the code so it passes a dictionary instead.

jamestwebber commented 2 years ago

Great news 🙃 the models will run but the results are very slightly different across python versions. Not sure why.