Alloy terminology and examples

Folknology 2020-11-09 20:27:27 UTC #21

Was 'x' 1 bit shaped and not a python boolean?
try ~ on a python boolean in a repl

Folknology 2020-11-09 20:40:38 UTC #22

The whole nMigen/Python thing is an inconsistent contextual tripwire IMHO, I have had similar problems with other Python DSLs (you think you are still in Kansas but it depends..) it messes with your pattern recognition of the language.

lawrie 2020-11-09 20:55:52 UTC #23

There is this warning for some uses of ~ in If

github.com

nmigen/nmigen/blob/master/nmigen/hdl/dsl.py#L212-L221


def _check_signed_cond(self, cond):
    cond = Value.cast(cond)
    width, signed = cond.shape()
    if signed:
        warnings.warn("Signed values in If/Elif conditions usually result from inverting "
                      "Python booleans with ~, which leads to unexpected results. "
                      "Replace `~flag` with `not flag`. (If this is a false positive, "
                      "silence this warning with `m.If(x)` → `m.If(x.bool())`.)",
                      SyntaxWarning, stacklevel=4)
    return cond

That check is made for If but not for Mux. Perhaps it should also be done for Mux. But in the cases that gave the wrong results for Mux(~x, ...) for me, I wasn't (knowingly) using a python boolean.

I probably ought to investigate this and the Repl case more, but i just wanted to get something working.

BigEd 2020-11-10 07:52:09 UTC #24

Umm, yes, I see what you mean. Looks like the first checkin was 66 lines, but it's grown a bit. I think 11 or 12 lines fall away if you remove the conditional compilation, but it's still over the 66 lines. It's a big page.

I had another look at what revalidinho said... it's the assembly code which comes out cleaner, is perhaps what I missed. So, OPC5[LS] is a smaller machine, OPC7 is a nicer machine to program or have a compiler target, and OPC6 is somewhere in the middle.

But great work in condensing the OPC6 in nMigen down to a page!