prevalence of "right" time zones without timegm()?

[Tom Yu]

"Henry B. Hotz" <hotz at jpl.nasa.gov> writes:

> On Jul 5, 2011, at 4:20 PM, krbdev-request at mit.edu wrote:
>
>> Date: Tue, 05 Jul 2011 15:18:59 -0400
>> From: Tom Yu <tlyu at MIT.EDU>

>> I propose that we implement krb5int_gmt_mktime() using timegm() on
>> systems that have it, using our existing custom implementation if
>> timegm() doesn't exist.
>
> How does it compare to mktime()/tzset() with the right TZ?

Greg Hudson already mentioned the multithreading issues with that
approach.  Determining the "right" TZ to use in this situation is also
nontrivial.

> Or is this what you're putting in your custom krb5int_gmt_mktime()?

The existing custom krb5int_gmt_mktime() implements the equivalent of
timegm() that has no knowledge of leap seconds, i.e., every minute has
exactly 60 seconds.