Add ReadableStream support to k6

[joanlopez]

What?

It adds support for ReadableStream from Streams API through experimental k6/experimental/streams module.

Why?

This is a prerequisite to unlock the development of other features in k6. It is also a nice-to-have feature for k6 users in general, that may want to use ReadableStream on their own.

Note for the reviewer

You can find the rationale behind some technical decisions (around goja, promises, etc) present on this changeset, in this
Google Doc that I used to write some personal notes, as takeaways during the development process.

On top of that, @oleiade and I decided to keep the code comments for each step of those operations defined in the spec because it makes it easier to:

• correlate the code and the spec.
• identify bugs or non-compliant behaviors.
• identify patterns that are repeated over the spec, and its correspondence in Go/goja.

So, although I know these could be arguable for some people, I'd personally push to keep them as long as I need to maintain and take care of this code.

Checklist

• I have performed a self-review of my code.
• I have added tests for my changes.
• I have run linter locally (make lint) and all checks pass.
• I have run tests locally (make tests) and all tests pass.
• I have commented on my code, particularly in hard-to-understand areas.

Related PR(s)/Issue(s)

Closes #3659
Contributes to #2978

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 0.28249% with 706 lines in your changes are missing coverage. Please review.

Project coverage is 70.79%. Comparing base (f38ac59) to head (598ac70).

❗ Current head 598ac70 differs from pull request most recent head 89302f2. Consider uploading reports for the commit 89302f2 to get more accurate results

Files	Patch %	Lines
...odules/k6/experimental/streams/readable_streams.go	0.00%	208 Missing ⚠️
...ntal/streams/readable_stream_default_controller.go	0.00%	134 Missing ⚠️
js/modules/k6/experimental/streams/module.go	0.84%	117 Missing ⚠️
...rimental/streams/readable_stream_default_reader.go	0.00%	63 Missing ⚠️
.../k6/experimental/streams/readable_stream_reader.go	0.00%	59 Missing ⚠️
js/modules/k6/experimental/streams/goja.go	0.00%	44 Missing ⚠️
js/modules/k6/experimental/streams/queue.go	0.00%	25 Missing ⚠️
js/modules/k6/experimental/streams/errors.go	0.00%	24 Missing ⚠️
...dules/k6/experimental/streams/underlying_source.go	0.00%	17 Missing ⚠️
js/modules/k6/experimental/streams/errors_gen.go	0.00%	11 Missing ⚠️
... and 1 more

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3696      +/-   ##
==========================================
- Coverage   73.29%   70.79%   -2.51%     
==========================================
  Files         278      287       +9     
  Lines       20456    21159     +703     
==========================================
- Hits        14994    14979      -15     
- Misses       4505     5217     +712     
- Partials      957      963       +6     

Flag	Coverage Δ
ubuntu	70.79% <0.28%> (-2.44%)	⬇️
windows	?

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.

[joanlopez]

Hey @mstoykov (cc/ @oleiade),

I have pushed a few more commits with some changes that go in the line of what has been discussed here.

Note that the approach is specific for streams, thus not extensible, but I guess we can postpone the task of finding a way to make the WPTs setup available for other suites as a separate work.

I also took the opportunity to remove the support for console.log, as discussed here, in order to move that discussion outside the scope of this pull request, as it isn't strictly needed.

Please, let know what do you think about the new approach. Thanks! 🙇🏻

[mstoykov]

I guess here set_status above should be the thing that marks tests as failed?

[joanlopez]

Yeah, but that status isn't directly reflected in any way through how we do run the tests, just kept in-memory stored as information associated with the test.

That's why I suggested there to explore a mechanism to do so, like "building" our own test environment, but I'd prefer to leave that for the next iteration as for now it's more than enough (it reports the failures) and the major drawback is that it fail fast and doesn't run all the tests when there's an error, which shouldn't be that bad for an initial release, as long as the behavior is correct.

(cc/ @oleiade as you'll most likely take over 🙇🏻)

[mstoykov]

LGTM in general.

I have left some more nits.

I am still not certain about a bunch of the stuff around the running of the tests.

Also if locked is supposed to be read only - the test definitely do not test for this as having as an exported field on a struct let it be set from goja.

So maybe we should have more test than just hte ones in the test suite

[olegbespalov]

LGTM

[mstoykov]

Just please squash on merge 😬

[oleiade]

👍🏻 😄 🦖