Trevor Elliott

> I realized that #4484 hadn't merged yet and I had forgotten to do a review (and now it needs a rebase to merge) but once it merges, we should...

> Is it important to have a build-time flag for whether to generate this code? Or would it be okay to just always generate the `trace!` call? I assume this...

> > The reason for the flag was mostly that the tests aren't compiled in a context that has access to the `trace!` macro. > > Huh, I don't understand...

After playing around with this a bit, it's not actually that useful: the log message printed tells you that the top-level function you applied failed, which you already knew :)

I've finished up translating icmp for x64

I'm working on finishing the x64 migration to ISLE. Feel free to grab instructions if you'd like to work on them, otherwise I'll continue working down the list.

The x86_64 backend has been migrated to ISLE :tada:

That's a good point. The only downside that I can see there is that we'd need to be really careful that we were always introducing the `RegMem` variants when it...

I ran into another instance of this today: the `sse_cmp_op` rule in the x64 lowering has overlapping cases for vectored floating point types. The rules on lines on 1504 and...

It's worth pointing out that for the example in https://github.com/bytecodealliance/wasmtime/issues/4717#issuecomment-1217308075 we would still not be able to raise shadowing errors: the patterns that shadow the constant patterns are made up...