4.0 release plan

[krassowski]

This is the current plan for 4.0 release:

• move plugins to @jupyter-lsp namespace
• refactor jump-to function to suport jumping to references (Implement jump target selector and jump to references #739)

Postponed for 5.0 (#917):

• deprecate APIs which are included in JupyterLab's 4.0 @jupyterlab/lsp package
  • try to harmonise them to minimise the cost of switchover once lab 4.0 is released
  • remove previously deprecated methods
• turn on more eslint rules, including camelCase CamelCase naming convention? #47
• split up features into packages Reorganize client source with lerna and typescript projects #76
• add a scaffold for new features:
  • symbol navigation (sidebar panel, + modal) Implement navigation to symbol (workspace/symbol) #43
  • code actions and commands (How to invoke executable commands for an LSP server #676)
• typdoc documentation

If time permits:

• multiple language servers per document (will close Support multiple language servers per document #437)
• Manually re-launching LSP servers #681

[bollwyvl]

These are very good goals, thank you for picking up this part of the planning! I'll see how much time i can put towards assisting.

Regarding timeline/upstreams: are we thinking this will target jupyterlab 3 and retrolab, or should we be shooting more for lab4/notebook7?

If we do go after #437, I think we want to make sure it covers as much of #184 as we can. A win from that would be at least some js-based language servers running inside the browser. While this would need to work outside of jupyterlite. This would improve our demo capability (see binder issues on #724) and, for things like the JSON/YAML language servers, substantially improve our ability to deliver just-in-time features, shipped only in the browser, and without nodejs.

[krassowski]

Regarding timeline/upstreams: are we thinking this will target jupyterlab 3 and retrolab, or should we be shooting more for lab4/notebook7?

I would currently aim to release it with 3.3, but if it is not ready before Lab4 release then I guess it makes sense to add support for Lab4 too.

I was reading through #184 yesterday and I agree that these issues should be considered together. My first thought was that I would not want to land #184 before #437, so added #437 as a more realistic (time-wise) goal.