Bug#89098: Timezone POSIX compliance bug.

[Paul Eggert <eggert@CS.UCLA.EDU>]

Roland McGrath <roland@redhat.com> writes:

> Off hand I cannot see why we clip the given values at all.

I'd guess that a reasonable amount of glibc code will break if the UTC
offset is enormous (e.g., on the order of 2**31 seconds).  Also, I
wouldn't be surprised if some of the time-zone-adjusting code (e.g.,
tzset, mktime) assumes relatively-sane UTC offsets, on the order of
days rather than years.

I suspect that these assumptions won't dump core if violated
(i.e., you'll get unspecified results, not undefined behavior)
but it'd take a reasonable amount of work to verify this suspicion.