time in Argentina

[Paul Eggert]

   Date: Wed, 19 Apr 2000 17:31:44 +1000
   From: Alex LIVINGSTON <alex at agsm.edu.au>

   This would have to be a classic case of multiple negations combined with
   mind-boggling ambiguity and perhaps translation difficulties.

Yes.

My understanding is that the Argentina has not adjusted its wall
clocks since 1993, and does not currently plan to adjust them in the
future.  However, Argentina did observe DST last summer because it
subtracted an hour from the standard-time offset during the period
that it was observing DST.  That is, ARST was 3 hours behind UTC in
1999/2000, even though it was 2 hours behind UTC in 1992/1993.  The
transitions starting with 1992/1993 are as follows:

Sun Oct 18 02:59:59 1992 UTC = Sat Oct 17 23:59:59 1992 ART isdst=0
Sun Oct 18 03:00:00 1992 UTC = Sun Oct 18 01:00:00 1992 ARST isdst=1
Sun Mar  7 01:59:59 1993 UTC = Sat Mar  6 23:59:59 1993 ARST isdst=1
Sun Mar  7 02:00:00 1993 UTC = Sat Mar  6 23:00:00 1993 ART isdst=0

Sun Oct  3 02:59:59 1999 UTC = Sat Oct  2 23:59:59 1999 ART isdst=0
Sun Oct  3 03:00:00 1999 UTC = Sun Oct  3 00:00:00 1999 ARST isdst=1
Fri Mar  3 02:59:59 2000 UTC = Thu Mar  2 23:59:59 2000 ARST isdst=1
Fri Mar  3 03:00:00 2000 UTC = Fri Mar  3 00:00:00 2000 ART isdst=0

This understanding is reflected in my latest proposed patch to tzcode.
If it's incorrect, please let me know, ideally by citing authoritative
sources.