On Tue, Aug 22, 2017 at 5:15 PM, Gregory Ewing <greg.ew...@canterbury.ac.nz> wrote: > Chris Angelico wrote: >> >> a naive ASCII upper-casing wouldn't produce 0x81 either - if it did, it >> would also convert 0x21 ("!") into 0x01 (SOH, a control character). So >> this one's still a mystery. > > > It's unlikely that even a naive ascii upper/lower casing algorithm > would be *that* naive; it would have to check that the character > appeared to be a letter before changing it. > > You might expect bytes >= 0x80 to be classed as non-letters by > that test, but what if it ignores the top bit or assumes it's > a parity bit to be left alone? What do you get under those > assumptions?
Exactly, I do assume that it's checking for it to be a letter. But everything previously has been on the assumption that it ignores the top bit. That's how we got this far. ChrisA -- https://mail.python.org/mailman/listinfo/python-list