Hi Oleg,
Thanks for discovering the problem. Mathematically equivalent Faust code
should, as much as possible, result in the exact same generated code. It is
clearly not the case here. We will take a look…
Cheers,
Yann
Le dim. 8 oct. 2023 à 18:52, Oleg Nesterov a écrit :
> Hello,
>
> simp
sorry for the noise, forgot to mention ...
Granted, I think that this special "ksub && izZero" case in simplification()
should die and I would be happy to make a PR.
But perhaps you can explain why is it useful or necessary?
Thanks,
On 10/08, Oleg Nesterov wrote:
>
> Hello,
>
> simplification()
Hello,
simplification() does
...
else if (isSigBinOp(sig, &opnum, t1, t2)) {
BinOp* op = gBinOpTable[opnum];
Node n1 = t1->node();
Node n2 = t2->node();
...
else if (opnum == kSub && isZero(n1))
return sigBinOp(kMul, sigInt(-1), t
