fix: lint files and create a global cache

[ricardogobbosouza]

This PR contains a:

• bugfix
• new feature
• code refactor
• test update
• typo fix
• metadata update

Motivation / Use-Case

Linting the necessary files and create a more intelligent cache of the result

Resolves #75 , #74

Breaking Changes

Additional Info

Thanks @watjurk

[codecov]

Codecov Report

Merging #79 (a696151) into master (f6d26b7) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master       #79   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            8         8           
  Lines          250       262   +12     
  Branches        70        74    +4     
=========================================
+ Hits           250       262   +12     

Impacted Files	Coverage Δ
src/index.js	100.00% <100.00%> (ø)
src/linter.js	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f6d26b7...a696151. Read the comment docs.

[ricardogobbosouza]

/cc @alexander-akait

Can use the invalid hook to invalidate cache?
Apparently it worked well

WDTY?

[alexander-akait]

invalid for watching, better don't use this hook, it will have same bad side effects with plugins and watching/serve

[ricardogobbosouza]

Can we keep it that way in v2 and fix it in v3? If not, we will have performance problems!

Or do you have any idea how to do this?

[alexander-akait]

I need to look at this issue, I would prefer to return to it in 3 versions, is it not critical now?

[watjurk]

Unfortunately I think this is critical.

it will have same bad side effects with plugins and watching/serve

what side effects ?

[alexander-akait]

Because some projects use manually invalidate for serve, need investigate how it can break this plugin

[watjurk]

@alexander-akait could you provide more details on how this can break this plugin, I'll be happy to investigate.

To resolve issue #83 i would suggest using both compilation.hooks.finishModules and compilation.hooks.succeedModule in following manner:

const seen = new Set()

compilation.hooks.finishModules.tap(ESLINT_PLUGIN, (modules) => {
  /** @type {string[]} */
  const files = [];

  // @ts-ignore
  for (const { resource } of modules) {
    if (resource) {
      const [file] = resource.split('?');

      if (
        file &&
        !seen.has(file) &&
        isMatch(file, wanted) &&
        !isMatch(file, exclude)
      ) {
        seen.add(file)
        files.push(file);
      }
    }
  }

  if (files.length > 0) {
    lint(files);
  }
});

compilation.hooks.succeedModule.tap(ESLINT_PLUGIN, (module) => {
  const { resource } = module
  const [file] = resource.split('?'); //? maybe move this to function
  if (
    file &&
    !seen.has(file) &&
    isMatch(file, wanted) &&
    !isMatch(file, exclude)
  ) {
    seen.add(file)
    lint(file)
  }
})

"Why don't use only compilation.hooks.succeedModule ?"
Because then (in watch mode) we would only lint files that have changed.
"Yes but then we could use crossRunResultStorage."
Then we would again face issue #74.

[jsg2021]

This cache held all results that had been computed... invalidations were supposed to be deleted from this and re-run.. so we don't recompute results for files that did not change... but still show them if they had issues.

[jsg2021]

Related #83

[jsg2021]

This feels like reinventing the results storage that was removed when all that was needed was to remove results for invalidated entries.

[jsg2021]

"Yes but then we could use crossRunResultStorage."
Then we would again face issue #74.

Just delete entries from it on invalidation events.

[watjurk]

@ricardogobbosouza can we get #79 (comment) into fix-lint-and-cache branch ?
@alexander-akait could you provide some details on how compiler.invalid could break this PR ?