feat(ext/web): Add reportError()

[nayeemrmn]

Blocked on #14159.
Closes #13484.

[bartlomieju]

@nayeemrmn I think this is good implementation wise. I want to get some more eyes on it before landing.

@crowlKats @lucacasonato please take a look

[lucacasonato]

Actually, WPTs don't pass with my WPT patch:

Going to run 2 test files.
----------------------------------------
/html/webappapis/scripting/reporterror.any.html

test self.reportError(1) ... ok
test self.reportError(TypeError) ... ok
test self.reportError(undefined) ... failed
test self.reportError() (without arguments) throws ... ok
test self.reportError() doesn't invoke getters ... ok
test stderr:

file result: failed. the event loop run out of tasks during the test

[nayeemrmn]

@lucacasonato #13799 (comment). Chrome also fails that right now. We can get compliant FireFox behaviour by adding an internal escape hatch to ErrorEvent. I'd like to either do that in a follow up PR or wait for resolution on the spec side in case they go with option 1.

[lucacasonato]

My concern is not that failing test, but the "event loop ran out of tasks" thing.

[nayeemrmn]

My concern is not that failing test, but the "event loop ran out of tasks" thing.

Running locally I got varied results... eventually I reproduced this but now it's happening for every test even after I switch wpt back to main branch 😓I'm going to try in CI here.

[nayeemrmn]

My concern is not that failing test, but the "event loop ran out of tasks" thing.

It seems like it's not related to this PR, but some change in wpt since web-platform-tests/wpt@96f98c4 (deno main) and wpt master is making this happen to every test.

[bartlomieju]

Discussed with @lucacasonato, he'll try to fix the WPT runner before merging this PR

[lucacasonato]

@nayeemrmn Can you rebase once more? WPT is updated on main, and should now be working once again.

[nayeemrmn]

@lucacasonato Ready

[lucacasonato]

LGTM!

[lucacasonato]

Sorry this one took so long to land @nayeemrmn. Thank you very much for this work!

[KnorpelSenf]

Given #13987 (comment), this seems to be the only other PR which affects networking in the 1.21.0 release. Could this be responsible for #14437? Based on the code changes, I don't really understand how, though …

[KnorpelSenf]

fetch is still not able to send files reliably. Does anyone know if this is actually the offending PR? Are there maybe any other changes in https://github.com/denoland/deno/releases/tag/v1.21.0 that could have broken it?

It seems to me like fetch is a relatively important thing for Deno and I'm a bit puzzled that one cannot trust it to send buffers and byte streams, as it seems to be messing with the data in some ways. In case you're not willing to look into this, could you shed some light on why #14437 isn't a priority?