Veikka Tuominen

> I think we could implement this by just truncating the argument at the first newline before writing it to `macro_buf` in `main.zig:parseArgs` Sounds good to me.

The type system especially is in a serious need of being redesigned but while using InternPool would give performance and memory usage improvements, it wouldn't directly solve the biggest pain...

> There's a rust crate that does this, which is MIT licensed, which could possibly be used as inspiration: https://github.com/mahkoh/repr-c/tree/master/repc/facade That does seem like the best place to start. >...

> I'm not sure what the plan for AROCC is regarding the intended C compiler (this is different from ABI?). One of the main goals of Aro is to replace...

> Mahkoh's rust layout seems to imply that clang and gcc layout bitfields differently in some circumstances. As seen in this code snippit : That does seem to imply something...

It'd probably be best to add a new file which does record layout calculations and call it from `recordSpec` in `Parser.zig` where there currently is a TODO comment.

I was thinking that we could check existing and new attributes for the largest alignment and only store one alignment attribute with that type. This deduplication should also happen to...

The same mechanism should also be able to detect use of uninitialized variables in general (in the limited scope of per function static analysis).

That'd be nice, I was planning on writing something like linenoise but never got around to do it.

I do think more different statements/control flow could be nice, but I'm not sure how useful repeat..until/do..while would be since I've never needed them.