Nice! I'll have a look through this later today.

🎆

Really looking forward to rewriting the C++11 backend on this baseline. Do you guys feel comfortable enough with what's here to recommend me working off of it? I do of course anticipate changes, but I'd also like to be able to provide feedback (for C++'s needs) at the right time.

I think so, but then I would say that as it's my work so far ;)

We've not really discussed where it's at yet, but I'm mostly happy with the CoreImp AST as it is here (except for potentially adding those cases for dealing with array length checking/index reading/slicing).

I'd really like to know how it works out for your case and if there are changes we need to make. At the moment it was mostly built with JS/Lua/JVM capabilities in mind, but only because that's what I'm most familiar with.

I'll rebase when I get a minute tonight so it's not lagging behind master.

Great, thanks. I'll at least start having a closer look, then. I'm currently reviewing (and possibly rethinking) how typeclasses are represented in C++, so this might in turn have an influence on that.

Typeclasses are desugared into regular data constructors now, so hopefully you won't need a special representation for them at all. That currently happens in the CoreFn -> CoreImp desugaring step, but we hope to move it way up the chain before typechecking even (#928 - the typechecker has some issues that prevent it currently).

@andyarvanitis Ok, this is in a more up-to-date state now.

Thanks!

Coverage decreased (-1.38%) to 77.13% when pulling 97c448b on core-imp into 879a030 on master.

Maybe we should shift this to 0.9 too.

This is the PR, right? The issue is in 0.9 already 😄

Doh 😄 I just looked at the milestone and this popped up.