Agenda for sync meeting 10/2/2020

[tlively]

The next sync meeting will be Friday, October 2 at 9:00AM - 10:00AM PDT/ 6:00PM - 7:00PM CEST). Please respond with agenda items you would like to discuss. If no agenda items are posted here 24 hours before the meeting, we will cancel.

If this meeting doesn't already appear on your calendar, or you are a new attendee, please fill out this form to attend.

[abrown]

Hopefully we can resolve the dot product discussion:

• with more data, we may be able to decide whether to merge i32x4.dot_i16x8_s instruction #127 or label it as post-mvp
• from the discussion today it would seem as though #328 should be labeled post-mvp

[Maratyszcza]

I'd like that we decide on instruction proposals with experimental data:

• i32x4.dot_i16x8_s and i32x4.dot_i16x8_add_s instructions (i32x4.dot_i16x8_s instruction #127)
• v128.load32_zero and v128.load64_zero instructions (v128.load32_zero and v128.load64_zero instructions #237)

and decide on removing integer Not Equals instructions (#351)

[ngzhian]

Discuss #256, which makes out-of-bounds lane index immediate for instructions like replace_lane and extract_lane a parse/decode error (malformed) instead of validation error (which they currently are).

[tlively]

It would also be good to have a wider discussion of what to do about feature detection: #356

[lars-t-hansen]

Probably also whether v128 is a number type (#363), as it affects how the SIMD proposal fits into the larger language.

[tlively]

Thanks everyone! We didn't get to cover everything, but we will have our next meeting in two weeks. The calendar event should be updated, so let me know if something went wrong there. The notes from this meeting are at: https://docs.google.com/document/d/1JmdwVuQsTmP_sSwti0AV_wSFIc9Y2s8vIIeBIP8kEZ8/edit?usp=sharing