new integer type just dropped: singed int
@cfbolz @joe goes well with the venerable char
@jrose @joe now that you say it, this makes it totally obvious that char is a singed integer type
@cfbolz @jrose @joe i8 a singed char
@cfbolz @jrose @joe I mean, I just took a byte
@regehr @jrose @joe when we managed to compile pypy to C for the first time in 2005, the resulting binary immediately crashed: our compiler assumed that chars are unsinged, GCC disagreed 🤦‍♀️
@cfbolz @regehr @jrose @joe arm was classically one of the rare unsigned char platforms but i think for 64 bit they switched to signed char for compatibility with the incorrect mainstream platforms
@fanf @cfbolz @regehr @jrose there were a few DOS/Windows C implementations where `char` was unsigned as well, iirc
@cfbolz @regehr @jrose @joe Upon running my fast Python interpreter on a PPC604, I also learned that default signedness on gcc was different on different architectures...
@sbrunthaler @cfbolz @regehr @jrose and let’s not forget the classic PowerPC four byte `bool`

@regehr That won’t compile on machines located north of 66° latitude… one needs the `arctic` qualifier in such cases. The resulting object code might look fishy, however.

@cfbolz @jrose @joe

@cfbolz @adamchainz I've been burned by those a time or two
@cfbolz
Finally! I've been telling the compiler I want that for years!
@brouhaha @cfbolz when do we get a singing int, that's what I wanna know

singing int freddieMercury;

@cfbolz unsigned int for low-security data

signed int if you need to be sure it hasn’t been tampered with

@cfbolz … are we saying “singed” like a musical note, or “singed” like gently burnt.
@mgaudet I was thinking of the latter. But clearly we are just freely associating at this point