Agenda for subgroup meeting - September 7

[tlively]

The next subgroup meeting will be on September 7, 2021 at 9-10am PDT (6-7pm CEST). Note that this is the day after Labor Day in the US. As with other Wasm meetings, we'll host this on Zoom.

No registration required if you attended a previous meeting. Fill out the form here if it's your first time: https://forms.gle/JehrAB4gWbtHjybt9.

Please reply with topics for the agenda.

So far we have @littledan and @takikawa presenting on requirements and the design space for the JS API.

[RossTate]

I would find it helpful if we were to have a discussion of what we're looking for (if anything) with respect to separate compilation for the MVP (as opposed to the Post-MVP), similar to the discussion we had on privacy/encapsulation. For example, we have already had discussions concluding that we expect mutually recursive classes to have to be compiled together in the same module, but do we also expect a class to always be compiled in the same module as its superclass? It affects the design/prioritization space, and I have gotten mixed messages on what we want. As another example, @askeksa-google uses a global dispatch table for method dispatch, rather than a v-table or i-table or im-table. That technique has the advantage of using call_indirect, which I understand benefits from optimizations not applicable to call_ref or the like, but has the disadvantage that it's not compatible with separate compilation. So knowing what is and is not in scope for the MVP regarding separate compilation can have a pretty big impact on the design/implementation space (and the turnaround time), and I would like to know what I should be taking into consideration at present.

[tlively]

I agree it would be helpful to have more clarity there. IMO the most useful framing of the question would be choosing a point on a spectrum of increasing expressivity/capability similar to how @lukewagner framed the question of what capabilities the component model should have. I strongly suspect that our answer will be equivalent to "whatever the current proposal supports," but it would be good to explicitly acknowledge what capabilities and linking models that includes as well as what that excludes. Would you want to lead that discussion, @RossTate?

[RossTate]

Sure, I can do that. I don't have a strong opinion myself for the MVP. I just want us to be aware of what the options are and entail and then explicitly agree on something and be coordinated.

[tlively]

Thanks @takikawa and @littledan! And thank you also to @RossTate for being willing to push our other agenda item to next time. The notes are here: WebAssembly/meetings#875.