Follow symlinks outside of watched folder

[ArjhanToteck]

Solves #231 and updates the README to match. Next.js depends on this package, and this would make development much nicer (vercel/next.js#53175).

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: ArjhanToteck (c54fa9f, 0cd204b, a5f52e9, 409ef51, 5b22101, 8aafd4a, 289ce5a, c54bf67, 696a67a)

[alexander-akait]

Looks like tests are failed, can you fix it?

[ArjhanToteck]

Still got a lot of remote environments failing the tests, but the machines I've tested on have been passing 100% of the unit tests reliably. I'll see if I can reproduce the failures somehow, unless it's not necessary as the required ones are passing.

[ArjhanToteck]

It's always the polling servers failing and I don't know why.

[alexander-akait]

         should detect a change to the direct symlink:
     Error: Timeout of 10000ms exceeded. For async tests and hooks, ensure "done()" is called; if returning a Promise, ensure it resolves. (/home/vsts/work/1/s/test/Watchpack.js)
      at Test.Runnable._timeoutError (/home/vsts/work/1/s/node_modules/mocha/lib/runnable.js:440:10)
      at Timeout.<anonymous> (/home/vsts/work/1/s/node_modules/mocha/lib/runnable.js:251:24)
      at listOnTimeout (internal/timers.js:554:17)
      at processTimers (internal/timers.js:497:7)

  2) Watchpack
       "before each" hook:

      Uncaught AssertionError: expected Array [] to equal Array [ '/home/vsts/work/1/s/test/fixtures/link2' ] (at length, A has 0 and B has 1)
      + expected - actual

      -[]
      +[
      +  "/home/vsts/work/1/s/test/fixtures/link2"
      +]
      
      at Assertion.fail (/home/vsts/work/1/s/node_modules/should/lib/assertion.js:92:17)
      at Assertion.value (/home/vsts/work/1/s/node_modules/should/lib/assertion.js:164:19)
      at /home/vsts/work/1/s/test/Watchpack.js:1290:37
      at Watchpack.<anonymous> (/home/vsts/work/1/s/test/Watchpack.js:1261:6)
      at Watchpack.emit (events.js:314:20)
      at Watchpack._onTimeout (/home/vsts/work/1/s/lib/watchpack.js:9:22455)
      at listOnTimeout (internal/timers.js:554:17)
      at processTimers (internal/timers.js:497:7)

[alexander-akait]

Will it help?

[ArjhanToteck]

Well, I've already seen the logs, I'm just wondering if you know whats different about the polling servers vs the normal ones, as only the polling servers are reliably failing these tests.

[alexander-akait]

I did some changes, please rebase

[codecov]

Codecov Report

Attention: Patch coverage is 91.66667% with 2 lines in your changes are missing coverage. Please review.

Project coverage is 92.88%. Comparing base (585ebf9) to head (8aafd4a).

❗ Current head 8aafd4a differs from pull request most recent head 696a67a. Consider uploading reports for the commit 696a67a to get more accurate results

Files	Patch %	Lines
lib/DirectoryWatcher.js	91.66%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #232      +/-   ##
==========================================
+ Coverage   91.55%   92.88%   +1.32%     
==========================================
  Files           6        6              
  Lines        1042     1054      +12     
  Branches      250      255       +5     
==========================================
+ Hits          954      979      +25     
+ Misses         88       75      -13     

Flag	Coverage Δ
integration	92.88% <91.66%> (+1.32%)	⬆️

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

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[alexander-akait]

Weird... Need to to tests...

[ArjhanToteck]

Ok so the issue is with the WATCHPACK_POLLING env variable. I guess that's what was different about those "polling" servers. I'm guessing the initial parameter in doScan is probably what's screwing things up, I'll look into fixing it.

[alexander-akait]

Feel free to feedback

[alexander-akait]

I am afraid about perf here...

[ArjhanToteck]

Didn't end up being necessary anyway, the new solution is simpler

[ArjhanToteck]

Tests are stuck again, should work this time tho

[ArjhanToteck]

The only thing now is that Watchpack Errors will print for a moment when a followed symlink is invalid, which is the case for a few of the tests, however this seems like it should be the correct behavior anyway, unless we don't want to warn users at all when an invalid symlink is watched.

[ArjhanToteck]

Ok looks like Windows is still failing when polling actually

[alexander-akait]

Do we need to do it in the sync way?