> I don’t want to be the killjoy but I find it highly questionable to
> add a function that is advertised as "secure" while we can't fully
> grok the complexities at play. If we can't produce a provable secure
> one, we should scrub the function for good; or at least rename it
> somehow.

The function is probably secure (modulo unseen bugs) in the
bytestrings-of-the-same-size case. To make it "provably" secure, we
could write a C version (which would be quite easy).

For unicode strings things are a bit trickier though. Again, a C version
could provide some guarantees (and could raise an error if the passed
unicode strings use a different representation from each other).