We ran out of time to finish the discussion... let's continue the conversation
I got this date from our previous when2meet... let me know if your plans have changed. I understand that @Anthony Bullard , @Ayaz Hafiz , and @Joshua Warner we're not available at this time.
Video call link: https://meet.google.com/krc-vseo-hvb
Sounds good to me, I'll let you know if something changes
Sorry I missed it - are there notes anywhere I can read?
No, uh, well...
:rofl:
292 messages were moved from this topic to #compiler development > zig compiler - spike by Luke Boswell.
Here's the PDF slides we started discussing... we got a little sidetracked around slide 15 :smiley:
(deleted) -- the version 2 of the diagram was in the wrong order, see the correct version below
When Sam is back from his vacation :palm_tree: :beach: I'll try updating it to incorporate more of our discussion points along with the paired zig structures etc.
89 messages were moved from this topic to #compiler development > zig compiler - IRs and pass ordering by Brendan Hansknecht.
Updated compiler on a page diagram (the old one was in the wrong order)
Roc Compiler Stages 1-page v3.svg
I just wanted to post here and bump this thread in case you missed it. As always, anyone is welcome to join.
I intend the scope for this discussion to be our compiler development plans (let me know if I'm getting distracted on a random tangent).
@Sam Mohr has put a lot of work into the overall structure as a starting point, so I imagine we can go through that -- making a note for anything we want to talk more about or things that need further design work.
I would like to avoid digging too deep into specific technical details and focus on the higher level abstractions like project structure and testing strategy etc, and our plans to collaborate. Specifically, I think we should talk more about what we think needs to happen and why, and then who is interested in doing what (in the short term).
Having a meeting is a great way for someone like @Sam Mohr who has thought deeply about things, to give an explanation of our current thinking or concepts for the compiler pipeline, and any clarification helps give everyone context.
If we want to go deep on a particular design, I think it would be better to have that kind of discussion in a Zulip thread where people can really get their thoughts out, or maybe schedule a followup conversation -- unless it's something that really affects everything.
The process so far has been translating the minimal parts we need from the Rust implementation and following Ayaz's cor experiment and design. I expect there to be lots of places where we can improve things, particularly with the IR's -- so if you're interested in this kind of thing, about now (right before we start building things) is probably the best time to have the biggest impact on the design.
The priority is to get an end to end compiler working with just the most basic features (i.e. functions and strings etc).
So, I imagine the next steps will be filling in all the missing pieces to implement each of the stages, which will be much easier to collaborate on with a clear structure and interfaces.
Let me know if you'd like to contribute or need any help. I'm no expert on compilers or anything, but happy to help point you in a direction (hopefully a good one).
See y'all in 90 minutes? Great presentation & ideas last time, and I'm planning to be a fly on the wall again for this one.
Yeppers
Sorry for not making it y'all. My daughter is sick and has thrown my whole day off
Last updated: Jul 06 2025 at 12:14 UTC