Schuyler Eldridge

@dtzSiFive: having Chisel be able to add statements to do the connections inside the when (carefully and safely!) is the best outcome here. This, however, very much goes against the...

> Just wonder if possible to share the cases among firrtl, verilog. FileCheck has support for checking multiple prefixes which is commonly used to share across separate invocations of some...

Thanks for reaching out. Adding Fixed Point back to Chisel as either libraries that wrap `UInt`/`SInt` or as something more clever would work. We don't want to support these (or...

> Regarding the test-driven approach you suggest, what would be the exact "goal"? To have the emitted FIRRTL be equal to the MLIR FIRRTL (up to the Fixed type annotations)?...

More clever would push in the direction of supporting user-defined types in FIRRTL and exposing that to Chisel. Users would then be free to define any type they need and...

> All good, but curious/interested: what's the reason/challenges preventing having :#= as a (sequence of) "secret command"? It's due to current implementation. This can be changed. The problem is that...

I think this is stemming from this being recently allowed in CIRCT and this specific code path never being possible before. This used to not be a candidate for memory...

@dtzSiFive landed support for this on the CIRCT side in https://github.com/llvm/circt/commit/d4ab551d910800632cdde403b5834c326d8c76fd. If you use the FIRRTL exporter that it provides, you now get the behavior above. This should be available...

Thanks for opening this, @debs-sifive!

This is totally fair... with https://github.com/chipsalliance/chisel/commit/a8f2ce7fe08c8461d89d54a584d278b476471439 landed, we're much closer to having a single file that needs to be passed as opposed to .fir and annotations. We should merge all...