fix #7100: remove @theia/languages and monaco-languageclient

[akosyakov]

What it does

• fix remove dependency to monaco-languageclient #7100: remove @theia/languages and monaco-languageclient - It allows to consume Monaco directly without trying to run vscode-languageclient in the browser. One should contribute language smartness via VS Code extensions instead.
• It cannot be part of 1.4.0, at least requires 1.5.0

TODO:

• following services should be deprecated in 1.4.0, see deprecate potentially used APIs in preparation to remove monaco-languageclient #8139:
  • Language -> use Language from @theia/core/lib/browser/language-service
  • Languages -> use monaco.languages.register*, MonacoLanguages.registerWorkspaceSymbolProvider or LanguageService from @theia/core/lib/browser/language-service
  • Workspace -> use MonacoWorkspace
  • semantic highlighting -> VS Code extension after Upgrade to Monaco 0.20.0 #8010 and Activating extension rust-analyzer failed: vscode.languages.registerDocumentSemanticTokensProvider is not a function. #7608
  • call hierarchy -> VS Code extension
  • type hierarchy -> TypeHierarchyRegistry.register for now, later VS Code extension: Create a "TypeHierarchyProvider" api for visualising type hierarchies microsoft/vscode#15533
  • languageServer namespace -> use vscode-langaugeserver via VS Code extension
• fix tests

How to test

• workspace symbols provided by VS Code extensions should be available:
  • wait till ts is initialized after opening a file
  • trigger Go to Workspace Symbol... command
  • type some prefix, wait till symbols are found
  • select some symbol and navigate to it
• suggestion in the debug console should be properly provided and applied
  • put a breakpoint in uri.spec.ts
  • debug with Run Mocha Test
  • use the debug console to lookup variables
  • use different prefixes, apply suggestions and check that they resolve to proper variable names
• language status bar entry should work
  • open some file
  • try to change a language
  • try to auto detect
• git blame hover should work
  • open some file
  • trigger blame annotations
  • hover over them and check that hover corresponds to commit in the ruler
• markdown preview widget should be in sync with text editors as before
  • open several markdown files and preview them (make sure to use preview from Theia, not built-in extension, one with old icon)
  • check that scrolling between preview and editor in sync
  • check that dirty changes are reflected
  • check that closed dirty editor without save reverts dirty changes in the preview
• call hierarchies contributed by VS Code extensions should work
  • open ts file and wait till ts server is ready
  • trigger Open Call Hierarchy on some method
• problem markers should be properly reflected
  • open some ts file
  • break it, should see markers in the problems view
  • delete it, markers should be removed
  • revert it in the scm view, markers should be added back
  • revert the dirty changes, markers should be removed

Review checklist

• as an author, I have thoroughly tested my changes and carefully followed the review guidelines

Reminder for reviewers

• as a reviewer, I agree to behave in accordance with the review guidelines

[akosyakov]

@tsmaeder @benoitf How do we deal with languageServer namespace? We don't want to add a dependency to a concrete version of vscode-languageserver in the plugin system. Should it be the same use VS Code extension instead? Do you use it in practice?

[tsmaeder]

@akosyakov I'm not aware of anyone using the languageserver namespace. AFAIC go ahead and kill it.

[akosyakov]

It's ready, but I don't think we need to review it yet. More important now to help with landing #8010 and fixing #7608 in this month. Otherwise everything will be postponed for one month.

[akosyakov]

it should be removed

[azatsarynnyy]

Tested according to How to test section.
Used GitLens extension for checking Git blame annotations.
Everything works as expected.

[RomanNikitenko]

I rebased and fixed conflicts for the PR as was discussed #8010 (comment)

There is copy branch without rebasing just in case https://github.com/eclipse-theia/theia/tree/ak/drop_monaco_language_client_copy

I'm going to wait for ci build/tests and provide light manual testing tomorrow morning before merge the PR.

[RomanNikitenko]

I did light testing after rebasing:

• workspace symbols
• debug console
• git blame hover
• call hierarchies
• problem markers

It works well for me - I don't see a regression.

So, I'm going to merge the PR and then adapt #8010 to these changes.

[kittaakos]

So, I'm going to merge the PR

Can we wait a few more days, please?

[benoitf]

hi @kittaakos, could you explain why you need more days ?

We just have a fresh tagged release so I think it's ok to merge all major changes in repo after this tag

like drop monaco-language-client, update FS api, bump monaco, etc.

[kittaakos]

hi @kittaakos, could you explain why you need more days ?

All downstream projects still using language contributions from Theia cannot use the next version anymore.

[benoitf]

@kittaakos I'm not sure few more days would help then ? as basically at some point next will still remove the support of monaco-language-client as target is for 1.5.0
I mean, if it's dropped on monday, it will still break on monday.

[kittaakos]

I mean, if it's dropped on monday, it will still break on monday.

I still have time to code until Monday.

[benoitf]

ok so let's wait until monday @RomanNikitenko ?

[RomanNikitenko]

I would like to land #8010 ASAP
but, It's OK for me to wait until Monday if it helps to resolve the problem for downstream projects

[kittaakos]

to resolve the problem for downstream projects

Yes, it helps. For example, here is a Theia issue: #8281
The only way to pick up such a fix by downstreams is by keeping the Theia version on next.

[benoitf]

@kittaakos we could do some 1.4 bugfixes release as well (we did some of them in the past)

[kittaakos]

I would like to land #8010 ASAP
but, It's OK for me to wait until Monday

@RomanNikitenko, go for it. Thanks for your patience.