pip install jupytext[myst] should install myst-parser==0.8.x

[mwouts]

Quick fix for #589

[codecov]

Codecov Report

Merging #590 into master will increase coverage by 0.02%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master     #590      +/-   ##
==========================================
+ Coverage   98.99%   99.02%   +0.02%     
==========================================
  Files          90       90              
  Lines        8699     8883     +184     
==========================================
+ Hits         8612     8796     +184     
  Misses         87       87              

Impacted Files	Coverage Δ
tests/conftest.py	100.00% <0.00%> (ø)
tests/test_cli.py	100.00% <0.00%> (ø)
tests/test_pep8.py	100.00% <0.00%> (ø)
tests/test_black.py	100.00% <0.00%> (ø)
tests/test_isort.py	100.00% <0.00%> (ø)
tests/test_config.py	100.00% <0.00%> (ø)
tests/test_mirror.py	100.00% <0.00%> (ø)
tests/test_combine.py	100.00% <0.00%> (ø)
tests/test_compare.py	100.00% <0.00%> (ø)
tests/test_execute.py	100.00% <0.00%> (ø)
... and 26 more

---

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 213acee...5f292f2. Read the comment docs.

[matthewfeickert]

This isn't really relevant, but if you wanted to have kept the compatible version syntax ~= this

jupytext/setup.py

Lines 57 to 58 in aeb2cf6

	extras_require={
	"myst": ["myst-parser>=0.8; python_version >= '3.6'", "myst-parser<0.9"],

could have just been

     "myst": ["myst-parser~=0.8.0; python_version >= '3.6'"], 

as that bounds to v0.8.X releases.

[mwouts]

Thanks @matthewfeickert . Well, before my commit we had something very similar to what you propose:

jupytext/setup.py

Line 58 in bca9173

"myst": ["myst-parser~=0.8; python_version >= '3.6'"],

However, I was quite surprise to see that it was not working so well. Not only LGTM was not understanding our requirement, but also, locally in my tests, pip install jupytext[myst] was installing myst-parser==0.12.x instead of 0.8.x! Is this something you can reproduce (or not)? Should we report this?

[matthewfeickert]

Well, before my commit we had something very similar to what you propose:

jupytext/setup.py

Line 58 in bca9173

"myst": ["myst-parser~=0.8; python_version >= '3.6'"],

Is this something you can reproduce (or not)? Should we report this?

I think this is just a misunderstanding of the syntax:

myst-parser~=0.8

means restrict myst-parser to v0.X where X >= 8, where you wanted to restrict to the v0.8.X range so you need myst-parser~=0.8.0. This is covered (the best that I've found) in PEP 440's "Compatible release" section

For a given release identifier V.N, the compatible release clause is approximately equivalent to the pair of comparison clauses:

>= V.N, == V.*

As an example, in a fresh virtual environment

(myst-example) $ pip install -qq "myst-parser~=0.8"
(myst-example) $ pip list | grep myst
myst-parser                   0.12.5
(myst-example) $ pip install --upgrade -qq "myst-parser~=0.8.0"
(myst-example) $ pip list | grep myst
myst-parser                   0.8.2

My understanding is that the idea here is that you're wanting to establish a lower bound on a release that will provide the API you want at the level of the minor or patch release, operating under the assumption that if people are actually using semantic versioning then pinning ~=1.X will allow you to get new features in minor releases (along with bug fixes in patches) while keeping in the v1.0 API of a library while ~=1.2.X means that you need to limit yourself to the v1.2 release of the library but you don't mind getting further patches.

Though not all libraries actually do this, and you might need to treat minor releases as you would normally treat major (c.f. scikit-hep/pyhf#999 for an example).

Does that make sense? Or am I misunderstanding what the problem is?

[mwouts]

Awesome! Thanks for taking the time to explain this! I'll use ~=0.8.0 then.