Adrian Sampson

Rad; this sounds a pretty legit change. In more than one way, it kinda reminds me of terminators in LLVM basic blocks… the decision there, however, is that the terminator...

Following up on @rachitnigam's call for deeper thinking on this in https://github.com/cucapra/calyx/pull/624#issuecomment-894636352: this is an intriguing issue, but I don't fully understand the implications yet. To summarize what I think...

Got it; thanks for the explanation! Perhaps stating the obvious again, but this sort of "pass-by-reference" thing needn't be exclusive to combinational situations… you can imagine the same setup being...

Huh, yeah. Enthusiastic yes to a broader conversation; this is a tricky issue. If the issue is mainly confined to the combinational case, there's a chance the underlying issue is...

Just copying a comment I made on Slack, for better organization: > I can imagine two options: > - Keep the mixed tree, and add some kind of symbol to...

@yn224, I'm moving discussion to #853, which is the issue about this specific problem.

Certainly time to re-evaluate, given all this progress! I checked off a few things—the stuff to be re-categorized (put on a roadmap somewhere, factored out into another issue, etc.) include...

Cool idea. And yeah, it could be special-cased just for `std_add`, for example, or it could rely on an attribute like `@shareable("add")` or whatever to indicate what "class" of shareable...

Very interesting! I like the idea of going for the more general `par`/`seq` rearrangement as a medium-term goal. At the end, though, you mention a drastically simplified version, which I...

No, #582 is the short-term "solution" where we use zeroes. This broader issue is about introducing an actual undefined state so we can throw errors when reading from undriven signals.