dkg at fifthhorseman
Mar 14, 2012, 3:45 PM
Post #3 of 4
(237 views)
Permalink
|
|
Re: 2.1.0beta3: make check fails on test 17 of common/t-timestuff when TZ=America/New_York
[In reply to]
|
|
On 03/14/2012 05:14 AM, Werner Koch wrote:
> On Wed, 14 Mar 2012 05:09, dkg [at] fifthhorseman said:
>
>> "make check" fails on test 17 (the last test) of common/t-timestuff
>> when TZ=America/New_York, but succeeds when TZ=UTC.
>
> Interesting:
>
> /* The date below is about the last time mktime works in CET on
> Windows XP; this is a somewhat strange because 32 bit Unices
> will happily work along for another month until they reach the
> end of all ticks on 20380119T031408 (unless Uli takes
> compassion on us and changes time_t to a u64). */
> { 2037, 12, 18, 23, 59, 59}
>
> The last bits of the Epoch will reach us earlier?
unlike tm_mday, tm_mon is zero-indexed (January = 0), so you're asking
for the 13th month of 2037, which is in fact Jan 2038, at the expiration
of the unix epoch (what the wall clock says when the epoch ends depends
on your TZ). I think the comment above is mistaken.
And yes, when the epoch rolls over, your wall clock will show a later
time than my wall clock (unless one of us moves before then). :)
--dkg
_______________________________________________
Gnupg-devel mailing list
Gnupg-devel [at] gnupg
http://lists.gnupg.org/mailman/listinfo/gnupg-devel
|
