polyfill-regenerator pollutes global scope even if set to "usage-pure"

[utrolig]

I might be misunderstanding here, but I thought using polyfill-regenerator was not supposed to pollute global scope if set to usage-pure. You can see a repro here.

When I build the app and open the index.html file, I can still see the regeneratorRuntime global variable.

[nicolo-ribaudo]

It's an error on my side, I didn't realize that regenerator always defines the global.

However, this is tracked at facebook/regenerator#302.

[stiboost]

Thanks for the fast response - does this mean that we would have to wait until it is resolved in the issue you linked, or are you able to resolve it?

[nicolo-ribaudo]

This needs to be fixed in regenerator itself, since currently it doesn't expose an entry point which exports regeneratorRuntime without also defining it as a global.

[utrolig]

So there is no way to completely polyfill and transpile the code down without requiring the regeneratorRuntime ?

[nicolo-ribaudo]

without requiring the regeneratorRuntime ?

I'm not 100% sure I understand what you are asking. If you look at your dist/bundle.js file, it never uses the regeneratorRuntime global variable.
It defines it (regeneratorRuntime=r, which is the same as window.regeneratorRuntime=r), but then never relies on it.

[utrolig]

Sorry I should've been more specific.

What I'm trying to achieve is to transpile my typescript application with async/await down to IE11 compatible code, without polluting the global scope. That includes polyfilling without scope pollution as well. There doesn't seem to be too much information about that on the web in general

[nicolo-ribaudo]

Thanks for clarifying.

Then the answer is "it's not possible until regenerator-runtime stops polluting the global scope (facebook/regenerator#302). When it stops doing so, this plugin won't need any change but you'll just need to update your regenerator-runtime dependency."

[juergba]

A new version v0.13.9 of regenerator-runtime has been released recently, including this commit.
Does it solve this issue?

[ljharb]

That commit would only impact CSP, not global scope.