It's a little bit more complicated than it seems from first glancing at the 'asia' file, because of the inclusion of the 'RussiaAsia' and 'EUAsia' rulesets.

`zdump` makes this somewhat clearer, though running the outputs through `diff` alone is only minimally helpful as all the transitions since 1992-09-27 are at least slightly different.  So, to help make clearer comparisons between proximal transitions, I sorted the outputs by date:

diff -U0 <(./zdump -i Asia/Baku) <(./zdump -i Asia/Yerevan) | sort -k 1.2

+++ /dev/fd/62 2019-07-21 12:43:51.000000000 -0400
+- - +0258 LMT
-- - +031924 LMT
--- /dev/fd/63 2019-07-21 12:43:51.000000000 -0400
-1924-05-01 23:40:36 +03
+1924-05-02 00:02 +03
+1992-09-27 02 +03
-1992-09-27 03 +04
+1993-03-28 03 +04 1
+1993-09-26 02 +03
+1994-03-27 03 +04 1
+1994-09-25 02 +03
+1995-03-26 03 +04 1
+1995-09-24 03 +04
-1996-03-31 06 +05 1
-1996-10-27 05 +04
+1997-03-30 03 +05 1
-1997-03-30 05 +05 1
+1997-10-26 02 +04
-1997-10-26 04 +04
+1998-03-29 03 +05 1
-1998-03-29 05 +05 1
+1998-10-25 02 +04
-1998-10-25 04 +04
+1999-03-28 03 +05 1
-1999-03-28 05 +05 1
+1999-10-31 02 +04
-1999-10-31 04 +04
+2000-03-26 03 +05 1
-2000-03-26 05 +05 1
+2000-10-29 02 +04
-2000-10-29 04 +04
+2001-03-25 03 +05 1
-2001-03-25 05 +05 1
+2001-10-28 02 +04
-2001-10-28 04 +04
+2002-03-31 03 +05 1
-2002-03-31 05 +05 1
+2002-10-27 02 +04
-2002-10-27 04 +04
+2003-03-30 03 +05 1
-2003-03-30 05 +05 1
+2003-10-26 02 +04
-2003-10-26 04 +04
+2004-03-28 03 +05 1
-2004-03-28 05 +05 1
+2004-10-31 02 +04
-2004-10-31 04 +04
+2005-03-27 03 +05 1
-2005-03-27 05 +05 1
+2005-10-30 02 +04
-2005-10-30 04 +04
+2006-03-26 03 +05 1
-2006-03-26 05 +05 1
+2006-10-29 02 +04
-2006-10-29 04 +04
+2007-03-25 03 +05 1
-2007-03-25 05 +05 1
+2007-10-28 02 +04
-2007-10-28 04 +04
+2008-03-30 03 +05 1
-2008-03-30 05 +05 1
+2008-10-26 02 +04
-2008-10-26 04 +04
+2009-03-29 03 +05 1
-2009-03-29 05 +05 1
+2009-10-25 02 +04
-2009-10-25 04 +04
+2010-03-28 03 +05 1
-2010-03-28 05 +05 1
+2010-10-31 02 +04
-2010-10-31 04 +04
+2011-03-27 03 +05 1
-2011-03-27 05 +05 1
+2011-10-30 02 +04
-2011-10-30 04 +04
-2012-03-25 05 +05 1
-2012-10-28 04 +04
-2013-03-31 05 +05 1
-2013-10-27 04 +04
-2014-03-30 05 +05 1
-2014-10-26 04 +04
-2015-03-29 05 +05 1
-2015-10-25 04 +04
@@ -2,3 +2,3 @@
@@ -29,41 +29,37 @@
-TZ="Asia/Baku"
+TZ="Asia/Yerevan"

As the last lines in the output above make reasonably clear, a '-' line indicates a line appearing only in the output for Asia/Baku (Azerbaijan) and '+' lines are for Asia/Yerevan (Armenia).

To me, there are five distinct periods that stand out:

(1) On 1992-09-27, Azerbaijan moved permanently forward to +04, while Armenia retained its ongoing seasonal shift from +03 to +04 through 1995-09-24.

(2) In 1996, the situation was somewhat flipped, with Armenia staying permanently on +04 while Azerbaijan employed a further seasonal shift to +05.

(3) Between 1997-03-30 and 2011-10-30, both used +04/+05, and transitioned to and from DST on the same dates, though at slightly different times — typically two hours apart, as Armenia would change at 02:00s (that is, 02:00 +04, or 22:00Z) while Azerbaijan would change at 04:00 +04 or 05:00 +05 (equivalently, 04:00s, or 00:00Z).  Given the ever-so-slight differences here, it would take some careful reference-checking to determine what exactly Nagorno-Karabakh did during this period.  However, with 30 transitions, it's certainly possible such sources could be found.

(4) From 2012 through 2015, only Azerbaijan continued to observe DST.  It could be interesting to determine whether the "enclave" went along with their Azerbaijani surroundings, or whether it followed Armenia and stayed on +04.

(5) From 2016 to present, both countries just observe year-round +04.  Nothing interesting there.

Of these, (1), (2), and (4) are of the greatest potential interest.  In particular, bits of (1) may be harder to verify since it is close to the fall of the Soviet Union.  (2) and (4) are at least similar situations.  (3) likely just boils down to minor technicalities which may wade further into the past and present politics of the region than is warranted, unless there happen to exist any clear authoritative sources which can be easily found.

--
Tim Parenti


On Sun, 21 Jul 2019 at 12:23, Artemis Tosini <tzdata@artemis.re> wrote:
From what I can tell in TZ data, the time zones only diverged between Asia/Baku and Asia/Yerevan in 1992.
Considering that the Republic of Artsakh declared independence in 1991, there is a chance that it has followed Armenian (Asia/Yerevan) rules since it declared independence, and thus there is no need for Asia/Stepanakert. However, this is just a guess, and we need more research before deciding whether there is a need to create the zone.

--
  Artemis Tosini




Am So, 21. Jul 2019, um 09:49, schrieb Alois Treindl:

Nagorno-Karabakh is an 'Armenian enclave' inside Azerbaijan, which seceded from Azerbaijan in the 1990s, and has some kind of unclear status of autonomy since a ceasefire agreement in 1994.

See https://en.wikipedia.org/wiki/Nagorno-Karabakh

and https://en.wikipedia.org/wiki/Nagorno-Karabakh_Autonomous_Oblast

Azerbaijan and Armenia have different time zone rules, represented in TZ by Asia/Baku and Asia/Yerevan.

Has anyone researched the time zone and DST rules in Nagorno-Karabakh?

If the time zone rules followed locally differ from Azerbaijan, the situation there justifies the creation of a separate TZ zone. The zone exists now since 25 years with rather high political stability.