Compress output for optional chain with multiple ?.

[nicolo-ribaudo]

Q	A
Fixed Issues?	Fixes #1, Fixes #2
Patch: Bug Fix?
Major: Breaking Change?
Minor: New Feature?
Tests Added + Pass?	Yes
Documentation PR Link
Any Dependency Changes?
License	MIT

• The first commit is shared with Fix transform of delete a?.b in function params #15739.
• The second commit optimizes transforming a?.b?.c from a == null ? void 0 : a.b == null ? void 0 : a.b.c to a == null || a.b == null ? void 0 : a.b.c, to only emit void 0 once.
• The third commit optimizes transforming delete a?.b from a == null ? true : delete a.b to a == null || delete a.b.
• The fourth commit optimizes transforming a()?.b?.c from var _tmp1, _tmp2; (_tmp1 == a()) == null || (_tmp2 = _tmp1.b) == null ? void 0 : _tmp2.c to var _tmp1; (_tmp1 == a()) == null || (_tmp1 = _tmp1.b) == null ? void 0 : _tmp1.c.

Terser doesn't do these optimization by itself, so this change also brings improvements to minified code.

[babel-bot]

Build successful! You can test your changes in the REPL here: https://babeljs.io/repl/build/54764/

[JLHwung]

The output is much better, thank you!

[nicolo-ribaudo]

@SimenB Multiple PRs are failing with this (flaky) failure, but I fail to see how it can be Babel-related. However, it only fails for our "breaking" tests and not for the Babel 7 tests.
Do you have any idea of what might cause that unlinking error?

[SimenB]

Hmmm, no. The test uses @babel/register tho, so there might be something to it.

I can update the test to use fs.rm with force instead - that should in theory be more tolerant to errors such as these and bubble up a better result? Or just ignore the failure if it doesn't matter.

2 times it fails to delete a file since it's not there, and once it fails to launch node since the file isn't there. Seems a bit odd... 😅

[SimenB]

See jestjs/jest#14296

[nicolo-ribaudo]

Thank you! I can test that PR in our CI to see if it makes the problem go away. Is it possible that those failure are just hiding a real error?

They started happening in #15736 🤔

[SimenB]

I can test that PR in our CI to see if it makes the problem go away.

That'd be great 👍

Is it possible that those failure are just hiding a real error?

That's what I was thinking