Zachary Yedidia

I'm going to move this over the to HWToLLVM pass to resolve the conflicts.

I think the CI is failing due to an unrelated bug that was in main at the time when it ran. Is it possible to re-run the test?

Great, thanks! I will make these additional changes, and I'm currently rebasing and resolving conflicts.

Thanks for the comments, I think I have made all the requested changes!

I removed the recursion from `canonicalizeBits`, so I think all the comments have been addressed, except the one regarding code duplication with FIRRTLFolds. I propose to just remove the new...

There is a test for subword assignment to a register. Bundles and Vecs have been lowered away by the time this pass runs, so I'm not sure if we need...

CIRCT just treats invalid values as 0, so this gets generated for your example: ``` module subword_2( input c, output [1:0] x); assign x = c ? 2'h2 : 2'h1;...

This PR doesn't modify the combinational cycle checker, so those examples cause errors. If you disable the check with `--firrtl-check-comb-cycles=false` then it generates this for the first circuit: ``` module...

Ah yes that's right.

Here is a simple example with a CHIRRTL memory: ``` circuit Ram : module Ram : input clock : Clock input reset : UInt input io : { addr :...