Drop corejs

[mstoykov]

Everything but:

• Promises that aren't really working
• define/lookup getter/setter which are AnnexB
• other stuff that aren't standard
  is now in goja, so there is no need for corejs

Things that will break:

• The workaround explained in this comment and on stackoverflow although this breaks also because of Disable some babel plugins that are now part of goja or are not useful #1822
• define/lookup getter/setter which are depricated. defineProperty is since ES5.1 at least, so it is very unlikely any code is using __defineGetter__ they were also added to the Ecmascript later in AnnexB so 🤷‍♂️
• From the rest almost everything is AnnexB stuff completely unrelated to k6 such as bold to make a bold HTML tag.
• or completely unsupported stuff that was at some point a proposal

The one thing that now going over this shim.js from the previous PR, which might give us trouble is that flatten was renamed to flat somewhere along the proposal process. But corejs had it with the old name as that happened after that version. I think we shouldn't do anything and see if this actually breaks for users as this is 1 line change and will align with what is in browsers for example where flatten isn't defined.

[codecov-io]

Codecov Report

Merging #1824 (5edd8d5) into chore/disableNotNeededBabel (37add28) will decrease coverage by 0.01%.
The diff coverage is n/a.

@@                       Coverage Diff                       @@
##           chore/disableNotNeededBabel    #1824      +/-   ##
===============================================================
- Coverage                        71.58%   71.57%   -0.02%     
===============================================================
  Files                              181      179       -2     
  Lines                            13939    13905      -34     
===============================================================
- Hits                              9978     9952      -26     
+ Misses                            3328     3323       -5     
+ Partials                           633      630       -3     

Flag	Coverage Δ
ubuntu	71.54% <ø> (-0.02%)	⬇️
windows	70.06% <ø> (-0.09%)	⬇️

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

Impacted Files	Coverage Δ
js/bundle.go	91.83% <ø> (+1.17%)	⬆️
lib/executor/vu_handle.go	95.23% <0.00%> (+1.90%)	⬆️

---

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 37add28...5edd8d5. Read the comment docs.

[imiric]

This is YUGE! 👏

Did you run some benchmarks before/after? It would be good to know the difference and mention it in the release notes.

[mstoykov]

Did you run some benchmarks before/after

I didn't have time for that as well ...

[imiric]

I'm OK with this, just two questions:

• There are a lot of new errors in breaking_test_errors.json. Are we OK with them? I can't really tell whether they are important or not.

• Maybe for a separate PR, but could you document how breaking_test_errors.json is generated exactly? I followed the js/tc39/README.md, emptied it with {} and ran the tests, but no JSON was generated, though I didn't wait for it to finish. Do you parse the out.log and convert it to JSON? Exact commands would be helpful here.

[mstoykov]

• all the new errors are define/lookup-getter/setters, which we decided to dro, and some Promise ones as again the tc39 tests aren't the best-tagged tests ever :(

• It is in the output. I run it with -v (not certain it's required 🤔 ) and write it to a file. Then I open it and search for FAIL - the JSON will be just before it. It takes 2-3 seconds to delete the file to only contain the JSON in vim and then I move it on the breaking_test_errors.json.

Ideally, this will be automated (somewhat) one day, but currently, if you don't clean the file it will not work properly and I don't want to spend an afternoon fixing that, given that we will likely stop touching this that much after a few more changes. If anything I am more interested in optimizing it to be kind of "faster" but that will take longer and given that it will naturally become faster, the less babel is needed, I am more for trying to

[na--]

LGTM. Added a breaking change tag, since we probably should at least mention this as a minor breaking change, despite the fact nobody will probably notice it. Please update the PR description with what you think the note in the release notes should be. Ideally now, while everything is still fresh in your mind.

[mstoykov]

@na-- I've updated the description