Closed ianhi closed 3 years ago
CI fails but I can confirm that the extension works.
But I think we should wait until the release of the final jupyterlab3 version to merge in order to keep compatibility with jupyterlab2 until the stable release of jupyterlab3.
Totally agree, but I knew how to do this and to avoid the gotcha of not knowing about ensureVimKeymap
so I just wanted to make this public sooner rather than later.
Once we decide on the package name, I will publish this to pypi.
Two things here if it ends up with a name like jupyterlab-vim
:
2\. Given that this is a pretty general name would you be open to giving a few other people write access/pypi access so the project is robust against any one person no longer being involved. (i.e. safegaurd against the jlab2 switch nightmare)?
Sure, I can set multiple maintainers for the pip repository.
* Though if jlab-vim makes it's way into core in 3.1 this would be less of an issue.
That would be the best solution IMHO, I hope it will be in core in the 3.1 version.
Two things here if it ends up with a name like
jupyterlab-vim
:1. @jwkvam should have an input (or ability to say "please don't")
So let's stick with jupyterlab_vim
?
I wonder what was so bad about nodejs 13? weird.
Looking at this more carefully: the build.yml workflow actually does everything that the pull-request action does already so maybe just delete the old pull-request script and rename the build one?
Also, we can set up automated releases to pypi - though the version I know how to use requires you to bump the version manually then you create a release using the github interface. Thoughts on disabling the automated version bumper and switching to script that publishes to both pypi and npm?
Also, we can set up automated releases to pypi - though the version I know how to use requires you to bump the version manually then you create a release using the github interface. Thoughts on disabling the automated version bumper and switching to script that publishes to both pypi and npm?
For Python, I know versioneer, that uses the tag of the commit. But we can bump it manually, there will probably not be many versions.
Actually it seems that the single source of truth for versioning is contained in the package.json. _version.py
just reads that file to get the version. So The automated version bump should still work, though it will need to be modified to also push to pypi
though it will need to be modified to also push to pypi
Yes, definitely. In the meantime, I pushed the package from this branch to pypi (https://pypi.org/project/jupyterlab-vim/). An hyphen was set instead of the underscore on the Pypi page, probably because the main folder has one, even if the package name has undescore (jupyterlab_vim-0.13.0-py3-none-any.whl). I sent you an invite for the pypi project, so you can push new versions as well.
Yes, definitely. In the meantime, I pushed the package from this branch to pypi (https://pypi.org/project/jupyterlab-vim/)
Nice! I'll edit my first post to reflect this.
I sent you an invite for the pypi project, so you can push new versions as well.
Awesome thanks. Though hopefully I'll never have to make use of that ability.
Hey @axelfahy heads up that jlab3 has been released now. Is there anything more we need to do here?
Hey @axelfahy heads up that jlab3 has been released now. Is there anything more we need to do here?
I don't think so, I will merge. Thank you!
Updates for jlab3. I took the opportunity to switch this over to the new style "federated" extensions that eliminate the need for a build step. A consequence of this is that a pypi package will need to be released and that the method for loading the vim keymap switched from importing it to calling the
ensureVimKeymap
function (more details in commit message).I also applied the recommended tsconfig, eslint, and prettier settings from the upgrade script and then linted in a separate commit.
Closes: https://github.com/axelfahy/jupyterlab-vim/issues/14
Installing now If anyone is realllllllly wanting this extension for jlab3 prior this being merged then you're in luck @axelfahy has uploaded this PR to pypi and you can install like so
History.md
briefly documenting the change.If this is a release, additionally do the following:
package.json
jlpm install
to updateyarn.lock