Node.js Loaders Team Meeting 2022-05-10 about loaders HOT 7 CLOSED

I think we need to cancel because no one is available to run the Zoom and stream to YouTube. @Trott or @aduh95 could we get @JakobJingleheimer added to those accounts?

from loaders.

Okay, I'll close this for now. @JakobJingleheimer do you mind opening the issue and requesting that?

from loaders.

Anything for the agenda?

from loaders.

Not sure - any opinion regarding #77 (comment) ?

Do you think, with the chaining being merged, it's worth discussing ambient loaders (and perhaps "should ambient loaders just be loaders", as I recall cspotcode suggested it and we didn't recall why it'd be best avoided)? I would like to try to make a prototype soonish, if there's consensus on the general design.

from loaders.

I would like to try to make a prototype soonish, if there’s consensus on the general design.

I think there was agreement on the general use case, that it might be nice to load loaders from https or from .zip files or wherever, and therefore we need loaders to affect subsequent loaders. I’m unsure if there’s any reason that that shouldn’t just always be the case; what abilities would we lose if all loaders behaved that way? https://github.com/danez/pirates exists to resolve conflicts between CommonJS loaders, so I feel like there are probably cases like those described in that project’s readme that would apply to ESM loaders?

So assuming there are reasons we need some loaders to affect following loaders and others to not do so, then we would need some way to signal such. I liked the idea of adding an export from the loader itself to express that; “ambient” is a little vague, so maybe something like export applyToFollowingLoaders = true (we can rename later). Since this would be opt-in, there shouldn’t be a risk of this new capability breaking any existing use cases.

If you’re eager to get coding, maybe branch off the latest node master, now that it has chaining, and see if you can add support for such an export? And we can fine-tune the design once we see the branch in action, assuming it’s not too much effort to get it working. Or we can have a meeting first to discuss the design further; I think we’ve pretty much discussed everything, but I will defer to others who think otherwise. If it turns out to be a mess to try to support this optional “apply to following” behavior, as in, the complexity involved is substantial and it would possibly work better to just have all loaders behave that way, then we should probably meet up and discuss what downsides (if any) that would entail.

I realized I can’t make next week, though, so you’ll either have to meet without me or we can schedule for the following week or later.

from loaders.

An export would not work because it happens after the file has been evaluated and that's too late: #63 (comment)

I sprained my wrist, so I don't have anything to discuss, but I can attend.

from loaders.

@Trott or @aduh95 could we get @JakobJingleheimer added to those accounts?

I also do not have access to the Zoom or YouTube accounts. @mhdawson might be able to help. I think the process is open something in the admin repo, and then after a few days with no objections and enough approvals, somebody somewhere makes the magic happen.

from loaders.