Naive implementation for #6097

[guybedford]

As in #6097 this adds System.register __moduleName reference support.

An optimization that can be included is to only add this argument when it is explicitly used in the code, making the output cleaner, but it would be removed by minification anyway if it is always output.

[msftclas]

Hi @guybedford, I'm your friendly neighborhood Microsoft Pull Request Bot (You can call me MSBOT). Thanks for your contribution!

This seems like a small (but important) contribution, so no Contribution License Agreement is required at this point. Real humans will now evaluate your PR.

TTYL, MSBOT;

[vladima]

the change per se looks good, however can you please also update test baselines so we can see overall impact?
For compiler tests this can be done via

# do a clean build and run tests
jake clean local runtests
# accept current test results as baselines
jake baseline-accept

Unittests baselines should be updated in code.

Also since this will introduce new intrinsic name that user cannot rename we probably should either add a collision detection logic to make sure that there will be no identifiers named __module in user code similar to what we do for _arguments and _this today or expose some way for a user to pick a name for this second parameter.

[guybedford]

@vladima sure, I've updated the baselines and pushed that through. If going the collision detection route it might be nice to also only add the reference when it is already a free global __moduleName reference in the existing code.

[vladima]

thanks! can you please also update baselines in transpile set suite: https://github.com/Microsoft/TypeScript/blob/master/tests/cases/unittests/transpile.ts ?

[guybedford]

Sure, no problem... it looks like the tests are passing now.

[vladima]

@guybedford great, thanks!

[DanielRosenwasser]

@vladima is that a sign off? Is this ready to go in?

[DanielRosenwasser]

Discussed offline; one current problem is that we'll never report an error if a user defines their own __moduleName. I'd prefer if we had a solution that protects the user, but I can't definitely commit until @mhegazy returns.

[guybedford]

There isn't any actual conflict - if a user chooses to redefine __moduleName they can. Better tracking of this variable use in the implementation is definitely something to consider though.

[mhegazy]

👍

[mhegazy]

thanks @guybedford!