Is %z broken for return values of time.gmtime()?

[Anssi Saari]

random832 at fastmail.us writes:

> I would argue that it _should_ be, and that it should populate it with 0
> in gmtime or either with timezone/altzone or by some sort of reverse
> calculation in localtime, but it is not. Another problem to add to my
> list of reasons for my recent python-ideas proposal.

Docs say that tm_gmtoff and tm_zone are available if the struct tm in
the underlying C library has it. Based on that I kinda expected Cygwin
to have it but apparently not.