feat: remove partial type-information program

[bradzacher]

BREAKING CHANGE:
[utils] sets program = null for getParserServices(context, true).
[typescript-estree] disallows errorOnTypeScriptSyntacticAndSemanticIssues if project not provided

PR Checklist

• Addresses an existing open issue: fixes Enhancement: Remove 2 year old backwards compatibility check from getParserServices #5508
• That issue was marked as accepting prs
• Steps in CONTRIBUTING.md were taken

Overview

A few years ago I experimented with leveraging TS's unused variable diagnostics to power a lint rule. As this diagnostic is "single-file", I added the ability for the parser to generate a program in "isolated" mode (i.e. no parserOptions.project required and types/diagnostics only produced for the single file) as well as the ability for plugins to interact with this program.

However this approach had some major drawbacks:

1. calculating diagnostics in TS is slow - so the lint rule was slow as there's no way to tell TS to calculate specific diagnostics.
2. the isolated types are completely useless because in most modules so many types are seeded from imports.

We never bothered going back to cleanup this work (until now).

This PR:

1. updates the parser so that it only works with ts.SourceFiles if we're not doing type-aware linting.
   • this should have the added bonus of speeding the parser up for non-type-aware runs as we're not going through the heavy process of creating a program!
2. updates the getParserServices util as appropriate
   • Note: I left the ability to consume parser services without type info because the node maps can be useful in some situations, or sometimes it's useful to consume compiler options if provided.
3. removes the semantic-diagnostics-enabled test because it's a test that we don't get any value out of - it's just emitting TS errors which might change from version to version
4. only allows errorOnTypeScriptSyntacticAndSemanticIssues if we have full type info.
   • This is because TS doesn't compute a lot of diagnostics without a program!

[typescript-eslint]

Thanks for the PR, @bradzacher!

typescript-eslint is a 100% community driven project, and we are incredibly grateful that you are contributing to that community.

The core maintainers work on this in their personal time, so please understand that it may not be possible for them to review your work immediately.

Thanks again!

---

🙏 Please, if you or your company is finding typescript-eslint valuable, help us sustain the project by sponsoring it transparently on https://opencollective.com/typescript-eslint.

[nx-cloud]

☁️ Nx Cloud Report

CI is running/has finished running commands for commit 526c1d1. As they complete they will appear below. Click to see the status, the terminal output, and the build insights.

📂 See all runs for this branch

---

✅ Successfully ran 47 targets

• Node 18 - nx test eslint-plugin
• Node 14 - nx test eslint-plugin --coverage=false
• nx run-many --target=lint --all --parallel
• Node 18 - nx test typescript-estree
• Node 14 - nx test typescript-estree --coverage=false
• Node 18 - nx test utils
• Node 18 - nx test visitor-keys
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx test type-utils
• nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 14 - nx test scope-manager --coverage=false
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• nx run-many --target=typecheck --all --parallel
• Node 18 - nx test scope-manager
• Node 14 - nx test eslint-plugin-internal --coverage=false
• Node 14 - nx test parser --coverage=false
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx test eslint-plugin-internal
• Node 18 - nx test eslint-plugin-tslint
• Node 14 - nx test ast-spec --coverage=false
• Node 18 - nx test parser
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx test ast-spec
• Node 14 - nx test utils --coverage=false
• Node 14 - nx test eslint-plugin-tslint --coverage=false
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 14 - nx test type-utils --coverage=false
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 14 - nx test visitor-keys --coverage=false
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 18 - nx run-many --target=build --all --parallel --exclude website --exclude website
• Node 14 - nx run-many --target=build --all --parallel --exclude website --exclude website
• nx run-many --target=build --all --parallel --exclude website --exclude website
• nx run-many --target=build --all --parallel --exclude website --exclude website

---

Sent with 💌 from NxCloud.

[codecov]

Codecov Report

Merging #6066 (526c1d1) into v6 (dde6861) will decrease coverage by 0.02%.
The diff coverage is 70.00%.

Additional details and impacted files

@@            Coverage Diff             @@
##               v6    #6066      +/-   ##
==========================================
- Coverage   91.66%   91.64%   -0.02%     
==========================================
  Files         354      354              
  Lines       12199    12200       +1     
  Branches     3581     3582       +1     
==========================================
- Hits        11182    11181       -1     
- Misses        721      722       +1     
- Partials      296      297       +1     

Flag	Coverage Δ
unittest	91.64% <70.00%> (-0.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
...-estree/src/create-program/createDefaultProgram.ts	78.26% <ø> (ø)
...estree/src/create-program/createIsolatedProgram.ts	64.00% <ø> (-12.00%)	⬇️
...-estree/src/create-program/createProjectProgram.ts	93.47% <ø> (ø)
...ges/typescript-estree/src/create-program/shared.ts	83.33% <ø> (ø)
...t-estree/src/create-program/useProvidedPrograms.ts	84.84% <ø> (ø)
...ckages/utils/src/eslint-utils/getParserServices.ts	10.00% <0.00%> (ø)
...escript-estree/src/semantic-or-syntactic-errors.ts	86.66% <75.00%> (ø)
...eslint-plugin/src/rules/consistent-type-exports.ts	97.70% <100.00%> (ø)
...kages/eslint-plugin/src/rules/naming-convention.ts	81.86% <100.00%> (ø)
...ript-estree/src/create-program/createSourceFile.ts	90.00% <100.00%> (+12.22%)	⬆️
... and 2 more

[JoshuaKGoldberg]

🔥 Mostly looks fantastic - just requesting changes on the // backwards compatibility check comment.

[JoshuaKGoldberg]

No remaining blockers, and a lot of goodness in this PR. Let's get this merged!! 🙌

[JoshuaKGoldberg]

Btw @bradzacher the imports all available rule modules unit test failure is fixed in the v6 branch.

[bradzacher]

For reference I just did a really quick perf test of the before-and-after for this.

$ TSESTREE_SINGLE_RUN=true node --cpu-prof ./node_modules/.bin/eslint --no-eslintrc --parser="@typescript-eslint/parser" --ext=".ts,.tsx" .

Loading the profiles into https://www.speedscope.app/ I can see that getProgramAndAST is about 100ms faster on our codebase.

It doesn't seem like much, but our codebase is relatively small - ~2.6k files for ~207k SLOC in total. For a larger code base I'd expect more of an impact, though it'll likely still be pretty small overall.

For reference the CPU profiles: PROFILES.zip