Re: fossil build failure on riscv64
Sure; I didn't mean in this case (where I've already switched to
sprintf in a debian patch), I meant in general. Especially in code
that needs to be performant, so you don't want to waste time
converting to 64 bits and back or special casing particular values.
Also, shouldn't user code not assume particular numeric
representations (like 2s-complement) just in case computers switch to
trinary or base -2 or something in the future?
It's pretty annoying that there are stdlib functions like abs(3) which
is int-to-int and is "undefined" for the most negative integer, but
there's no int-to-unsigned uabs().
Reply to: