Adrian Sampson

Hi, @0adb—sorry I somehow missed your very useful comment a few months ago! This was good thinking: to change the `phi` instruction's semantics to (a) require at least contiguous runs...

I believe this is finally resolved in the new SSA extension, as of #403.

FWIW, while I continue to dig into this a bit, here's a one-liner for detecting the problem: ``` verilator --lint-only --top-module wrapper =(fud2 minimized.futil --to verilog) ``` `=(...)` is a...

This one's a doozy!! Thanks for the compact reproduction. ## understanding the problem I tried to pull out the core of the problem in the lowered Calyx. The offending lines...

Thanks for the discussion. To distill the option space, one way to think about it is that we need to respect the Calyx requirement that every (non-combinational) component takes at...

Agreed; we should make the path templates reflect the newly converted file, not the old file. This currently doesn't happen because we run the encoding command with the destination path...

Aha, thanks! A discussion in another bug's thread: that's why I couldn't find it quickly.

@sdfg2: As a general rule, there's no news unless it's posted here. Please feel free to help out!

It is true that $format could be determined before the conversion while $bitrate cannot; good point! But it's also possible—and perhaps a simpler change—for us to just move the _entire_...

@Barracuda6: no, that's not this problem. The plugin's `auto` option puts the converted files in your music directory, which you have configured here: ```yaml directory: /music/in ``` The `dest` option...