contributor coordination meeting - Feb 2024 · contributing

Stream: contributing

Topic: contributor coordination meeting - Feb 2024

Luke Boswell (Jan 22 2025 at 23:54):

Here's a when2meet for another contributor coordination meeting.

We've got a lot of different things going on -- particularly with the evolution of the compiler internals and new features (like SD replacing Abilities) -- and I think it would be helpful to sync up on these things.

I'm hoping we can have a more detailed conversation around the plan to achieve static dispatch and how we manage the compiler's evolution to support new Can (caching etc), new Mono (fix lambda-sets) etc.

Is it feasible to keep the current pipeline fully intact and operational?

I'd love to put together a diagram which describes the future compiler pipeline at a high level, and loosely maps some of our current state to that. It might just be the way my brain works, but it's currently hard for me to understand all the things going on so I assume it's challenging for others too.

Norbert Hajagos (Jan 23 2025 at 09:01):

I've filled my availability, but I would be more of an observer, so if a core contributor has conflicting availability with me, I wouldn't want to "win" with the timing of the meeting. Because of this, I'm sending this printscreen in – as I don't know if when2meet shows my name or not for the organizer–, so that you can make more informed decisions. I'm interested in the discussion; would be cool to catch the meeting, if it happens to be possible :smile:
image.png

Luke Boswell (Jan 23 2025 at 09:26):

It does show your name :smiley:

Luke Boswell (Jan 27 2025 at 09:25):

Let's lock in 2025-02-01T08:00:00+11:00

Video call link: https://meet.google.com/tsh-fyar-tpd

Brendan Hansknecht (Jan 27 2025 at 16:46):

Guess I'll miss this one. Probably doesn't matter too much, I haven't actively worked on roc since the beginning of the month.

Sam Mohr (Jan 27 2025 at 16:49):

We'll update you after

Sam Mohr (Jan 27 2025 at 16:50):

We're mainly trying to align on strategy for getting the new pipeline off the ground

Luke Boswell (Jan 27 2025 at 19:57):

It was a toss between this weekend or next, and I figure it'd be nice to have the compiler planning chat sooner than later. It looks like we've only got a smaller window, like 30mins ... so I am thinking maybe we keep the discussion really focussed on just the compiler Can and Mono evolution, and potentially follow up with another meeting the weekend following where we've also got people available if everyone's keen.

Sam Mohr (Jan 27 2025 at 23:11):

Sounds good

Sam Mohr (Jan 29 2025 at 13:12):

For those that are curious about what we'll be going over: Luke and I are putting together a presentation to organize the discussion, but the supplementary material is this repo containing the roughly proposed IRs for the new compiler stages: https://github.com/smores56/proposed-roc-pipeline

Sam Mohr (Jan 29 2025 at 13:13):

You don't need to read through beforehand, but it will hopefully give everyone a picture of how the build phase of the compiler will work in the future, barring codegen.