JupyterLab 3 update

[martinRenou]

Add support for JupyterLab 3 extension system

[martinRenou]

Do you still really need Python 2.7?

[mwouts]

Awesome! Thank you @martinRenou . I'm very glad to see jupyter_packaging arriving here! I think it would make sense to bump Jupytext's version number to 1.8 to ship this update, what do you think? And yes I agree, we should drop Python 2.7, that will make maintainers and contributor's life simpler 😄

I'll prepare another PR for Version 1.8.0-dev, and drop Python 2.7, and then maybe we can rebase your PR on that? Thanks!

[martinRenou]

Actually, JupyterLab requires Python >=3.6 (I am sorry I see it just now), Python 3.5 reached its end of life a month ago (note that conda-forge does not build packages for 3.5 anymore).

Do you think we could drop this one as well?

[mwouts]

Haha.. let me think about that... I know many people who would be in favor of that (cc @chrisjsewell), and indeed I'd love to use f-strings, etc... I'll confirm tonight

[chrisjsewell]

Haha you know my feeling on that 😉

[mwouts]

Hi @martinRenou , I've started working on removing Python 3.5, but I'm not done yet.
The WIP is in this branch: https://github.com/mwouts/jupytext/tree/drop_python_35

I get an error that has been reported at other places: voila-dashboards/voila#728. Maybe you know how to fix it?

If that helps, feel free to rebase you PR on top of the drop_python_35 branch. Otherwise I'll try to fix that in the coming days, or next week if I can't make it easily.

[martinRenou]

I get an error that has been reported at other places: voila-dashboards/voila#728. Maybe you know how to fix it?

Yes, I've seen this issue many times in multiple places... Using latest jupyter_server might fix it? Otherwise installing pytest-tornado can fix it but it's really not ideal...

I am not in a hurry to get this PR in :) We will need to wait for JupyterLab to be released anyway, so I can wait.

[yuvipanda]

I'd love love to get this released ;)

[martinRenou]

I'll be back on this soon 😊

[mwouts]

Hi @martinRenou , tonight I'll ship the version 1.8.2 (with this PR: #700), hopefully this will end the subject of removing the support of Python 2.7 and 3.5. Maybe then we should just try to rebase your PR on top of that, and hope that the issue with jupyter_server will have vanished since our last attempt 😄

[martinRenou]

and hope that the issue with jupyter_server will have vanished since our last attempt

I think it should have vanished now :)

I'll rebase the PR tomorrow then or the day after tomorrow for sure, it's on my TODO list.

[mwouts]

I think it should have vanished now :)

I hope so! I've taken #700, feel free to rebase when time permits. See you @martinRenou 👋

[martinRenou]

Codecov is not running on the right commit Oo, I feel like it's lost because I force pushed to erase the old commit. Will try to close and reopen the PR, see how codecov behaves.

[codecov]

Codecov Report

Merging #683 (885c72b) into master (91223d2) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master     #683   +/-   ##
=======================================
  Coverage   99.12%   99.12%           
=======================================
  Files          95       95           
  Lines        9356     9356           
=======================================
  Hits         9274     9274           
  Misses         82       82           

Impacted Files	Coverage Δ
jupytext/__init__.py	100.00% <ø> (ø)
jupytext/version.py	100.00% <100.00%> (ø)

---

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 5f801bd...798dc2f. Read the comment docs.

[martinRenou]

I think we can ignore the red check, it's wrong.

@mwouts This PR is ready! :)

[mwouts]

Awesome! Thank you so much @martinRenou .

No problem with ignoring the red check codecov/projet/tests, its seems to be duplicated anyway (and one of the duplicates is green) 😄

I'll try to release this tonight, maybe in version 1.9.0 of Jupytext, would that be OK for you? Also I am wondering if we should increase (or remove?) the version of the lab extension, or even publish an update of the lab extension on npm - at least I'd like to update the documentation and let people know how to install a version of the extension that works with JupyterLab 2.0. Do you have a recommendation on these points?

[martinRenou]

My understanding is that the new labextension package will not be compatible with JupyterLab 2, so it's a breaking change for the labextension. But it won't be a breaking for jupytext (the Python package), so people will be able to install this new jupytext for JupyterLab 2 or 3, and for JupyterLab 2 they will still need to manually install the labextension.

Publishing the new package to npm might not be needed in your case. If you do it, I suppose JupyterLab 2 is smart enough to not pick it up, but I am not 100% sure.