Updating the plotly.js version used in the plotly mimetype renderer

[jonmmease]

Hi, plotly.py maintainer here. First off, thanks for supporting the plotly mimetype. This makes vscode-python a really nice environment for plotly.py users.

I'm working on a refactor/generalization of the plotly.py rendering subsystem (See plotly/plotly.py#1474) and one of the things I'd like to do is automatically apply a dark theme (https://medium.com/@plotlygraphs/introducing-plotly-py-theming-b644109ac9c7) when plotly.py detects that it's running in vscode. For this to work, the plotly.js version used in the vscode-python mimetype renderer would need to be updated so that it includes theme support.

I'd be happy to submit a PR for this if someone could point me to where the mimetype renderer is maintained (I didn't find it when searching this repo).

As a side question, do you have a recommendation for the best way to detect that a Python session is running in VSCode? Right now I'm looking for the presence of the VSCODE_PID environment variable.

Thanks! 🙇

Environment data

• VS Code version: Version 1.32.3 (1.32.3)
• Extension version (available under the Extensions sidebar): 2019.3.6139

[rchiodo]

Hi @jonmmease, thanks so much for the offer.

Our mimetype rendering is here:
https://github.com/Microsoft/vscode-python/blob/325e1f025565a4455cb2d3e8e23a53c60a47b20a/src/datascience-ui/history-react/cell.tsx#L278

We actually use nteract transforms to do our rendering of different mimetypes.

For your second question, which python session are you referring to? A general python process launched, or an iPython console hosted by Jupyter? We could probably add another environment variable when we start the Jupyter server if you need it for something.

[jonmmease]

We actually use nteract transforms to do our rendering of different mimetypes.

Thanks @rchiodo, that was helpful. Looks like the plotly.js version is being determined by the @nteract/plotly package (https://www.npmjs.com/package/@nteract/plotly), which is locked to version 1.0.0 from a couple of years ago:

https://github.com/Microsoft/vscode-python/blob/419ecb1f593b4c58af77a9230d1179e34aa4d5bb/package-lock.json#L1128-L1133

There is a 1.43.0 version available that was up to date as of a few months ago. so I think a good starting point would be bump @nteract/transform-plotly to version 5.0.0 which should cause @nteract/plotly to jump to 1.43.0.

Do you have any established processes for dependency updates like this? I'm happy to submit a PR with just the package.json/package-lock.json updates, but let me know if there's anything else that would make things easier on your end.

For your second question, which python session are you referring to? A general python process launched, or an iPython console hosted by Jupyter?

The latter, I think. Basically I would like to know when the Python session is likely to be displaying plotly figures in VSCode using the mimetype renderer. It is already possible to detect whether a Python session is an IPython session, so it might be accurate enough so simply combine this with looking for the VSCODE_PID environment variable.

[rchiodo]

Do you have any established processes for dependency updates like this? I'm happy to submit a PR with just the package.json/package-lock.json updates, but let me know if there's anything else that would make things easier on your end.

Submitting a PR would be the quickest/easiest I think. Our CI system for the PR should verify that the update works.

[olegtarasov]

Hi @jonmmease! I've ran into some issues trying to use latest plotly charts with VSCode's interactive window. Looks like those issues are caused by the old version of plotly bundled with VSCode Python extension. Are you going to make a pull request with version bump? I would much appreciate this :)

[jonmmease]

Hi @olegtarasov, I was hoping for nteract/minimal-plotly#12 to get merged and published first so that we could update to the latest version of plotly.js. I just followed up in nteract/minimal-plotly#13.

[captainsafia]

@jonmmease Thanks for following up on the other thread! I've merged and released the bump to @nteract/plotly. Moving forward, updates will be released on a weekly cadence alongside updates in the nteract monorepo. I hope this unblocks your work here.