Python: Add support for additional widget types (in viewer or plots pane)

[isabelizimm]

Once #1918 is merged, we will have basic support for the widgets such as ipywidgets, ipydatagrid, and ipyleaflet in the plot pane. Still not working ❌ is:

• %matplotlib widget mode
• bqplot
• ipycanvas
• pythreejs
• ipytree

Here is a gist of all the different widgets, for testing purposes

CC: @jgutman

[jgutman]

@isabelizimm I imagine these were all tested prior to the switchover to the later version of html-manager, it might be worth us re-testing to see whether there are any changes in what works/what doesn't after the update -- we should also specify here the module version of the above widget types that we tested, since we know now that makes a difference to whether it's compatible or not

[jgutman]

In order to support matplotlib widgets and others that require a live Python kernel, we need to enable an actual non-stubbed RendererContext.

See current TODO in

positron/src/vs/workbench/contrib/positronOutputWebview/browser/notebookOutputWebviewServiceImpl.ts

Lines 189 to 193 in aa6476a

	// Some features known to be NYI:
	// - Message passing between the renderer and the host (RenderContext)
	// - Extending another renderer (RenderContext)
	// - State management (RenderContext)
	// - Raw Uint8Array data and blobs

and stubbed RendererContext

positron/src/vs/workbench/contrib/positronOutputWebview/browser/notebookOutputWebviewServiceImpl.ts

Lines 208 to 219 in aa6476a

	const vscode = acquireVsCodeApi();
	import { activate } from "${rendererPath.toString()}"
	// A stub implementation of RendererContext
	var ctx = {
	workspace: {
	isTrusted: ${this._workspaceTrustManagementService.isWorkspaceTrusted()}
	}
	}
	// Activate the renderer and create the data object
	var renderer = activate(ctx);

Also see how this gets handled for ipywidgets in the jupyter extension
https://github.com/microsoft/vscode-jupyter/blob/4ca00b28cf3b890593e9c84d73b65e67f51a8061/src/webviews/webview-side/ipywidgets/renderer/index.ts#L244-L257

[peekxc]

Does this cover plotly / bokeh plots?

[juliasilge]

@peekxc I believe that plotly works:

Are you observing something different?

For bokeh, I do not believe we have that working correctly right now (the plot opens in an external browser).

[peekxc]

For bokeh, I do not believe we have that working correctly right now (the plot opens in an external browser).

Ahh, I was talking about bokeh; sorry, I assumed the behavior would be identical

[seeM]

#4172 and #4208 should get the following widgets working:

• anywidget
• bqplot
• drawdata
• ipydatagrid
• ipyleaflet
• ipympl
• ipytree
• ipywidgets – all widget types except Output which needs special handling
• pythreejs

There may be smaller issues with those, which we can log separately.

I haven't yet figured out why ipycanvas isn't working.

[testlabauto]

Verified Fixed

Positron Version(s) : 2024.08.0-24
OS Version          : OSX

Test scenario(s)

Validated with everything here: https://gist.github.com/isabelizimm/02f7ba16dc74b202935f6d3e9ed6b892
Everything looks good except ipycanvas, which is noted above

Link(s) to TestRail test cases run or created:
N/A