adapt to Python3.8 with PEP570

[stonebig]

.... please update also vendored joblib

[codecov]

Codecov Report

Merging #267 into master will decrease coverage by 0.31%.
The diff coverage is 33.33%.

@@ Coverage Diff @@## master #267 +/- ##
=========================================
- Coverage 88.22% 87.9% -0.32% 
=========================================
Files 1 1 Lines 552 554 +2 Branches 112 113 +1 =========================================
Hits 487 487 - Misses 42 43 +1 - Partials 23 24 +1

Impacted Files	Coverage Δ
cloudpickle/cloudpickle.py	87.9% <33.33%> (-0.32%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6ab5eed...d020f6e. Read the comment docs.

[stonebig]

if you want to be more precise, test should be "if sys.version_info > (3, 8, 0, 'alpha', 3):"

as in https://github.com/ipython/ipython/pull/11720/files#diff-1c766d4a0b1ea9ed8b2d14058b8234ab

[ogrisel]

This LGTM. I am not sure we really need to distinguish alpha releases though. Maybe sys.version_info >= (3, 8) is enough.

Any opinion on this change @pierreglaser?

[pierreglaser]

maybe hasattr(types.CodeType, "co_posonlyargcount")?

[pierreglaser]

This whole part of the code is only useful for versions of python before PEP 3104. For recent python version (python3?), cell_set(cell, value) could just be cell.cell_contents = value instead. +1 For merging this as is once it's ready, and dust this part of the code off later on.

[pierreglaser]

Can you add a test with a function that uses positional only arguments? You also need to change the reducer of code object (Would we have a CI against python 3.8.0.a4, I think every test involving code objects would have failed)

[ogrisel]

Closing in favor of #269.