Europe/Minsk: September 1992 transition is incorrect?
Hi there, When taking a look at 'zdump -v Europe/Minsk' with tzcode2016e/tzdata2016e running on a FreeBSD 10 system, I observe the following: Europe/Minsk Sat Sep 26 23:59:59 1992 UTC = Sun Sep 27 02:59:59 1992 EEST isdst=1 gmtoff=10800 Europe/Minsk Sun Sep 27 00:00:00 1992 UTC = Sun Sep 27 02:00:00 1992 EET isdst=0 gmtoff=7200 Now taking a look at the Zone definition of Minsk: 2:00 1:00 EEST 1992 Sep 27 0:00s 2:00 Russia EE%sT 2011 Mar 27 2:00s My question is, shouldn't this transition have taken place two hours earlier? Can anyone else reproduce this? -- Ed Schouten <ed@nuxi.nl> Nuxi, 's-Hertogenbosch, the Netherlands KvK-nr.: 62051717
On 30 June 2016 at 11:18, Ed Schouten <ed@nuxi.nl> wrote:
Can anyone else reproduce this?
I managed to reproduce the same behavior on one of my Debian virtual machines with tzdata 2016e-0+deb8u1 and zdump (Debian GLIBC 2.19-18+deb8u4) 2.19 I'm not presently seeing it in my personal development copy of the Git repository, running in Cygwin, but although I remade zdump, I'm not sure offhand if it's perhaps parsing an older tz binary. -- Tim Parenti
30.06.2016 в 17:18:29 +0200 Ed Schouten написал(а):
Hi there,
When taking a look at 'zdump -v Europe/Minsk' with tzcode2016e/tzdata2016e running on a FreeBSD 10 system, I observe the following:
Europe/Minsk Sat Sep 26 23:59:59 1992 UTC = Sun Sep 27 02:59:59 1992 EEST isdst=1 gmtoff=10800 Europe/Minsk Sun Sep 27 00:00:00 1992 UTC = Sun Sep 27 02:00:00 1992 EET isdst=0 gmtoff=7200
Now taking a look at the Zone definition of Minsk:
2:00 1:00 EEST 1992 Sep 27 0:00s 2:00 Russia EE%sT 2011 Mar 27 2:00s
My question is, shouldn't this transition have taken place two hours earlier? Can anyone else reproduce this?
Rules Russia were recently fixed to have transition from DST at 3:00 local time, so zdump's behaviour is expected. Wikipedia links to http://www.lawbelarus.com/repub/sub30/texf9611.htm (Act of the Cabinet of Ministers of the Republic of Belarus from 25.03.1992 N 157) which says clocks were to be moved forward at 2:00 on last Sunday of March and backward at 3:00 on last Sunday of September (the same as previous USSR and contemporary Russian regulations). So it looks like March 1992 transition for Europe/Minsk is incorrect. I suspect that lines 2:00 1:00 EEST 1991 Sep 29 2:00s 2:00 - EET 1992 Mar 29 0:00s 2:00 1:00 EEST 1992 Sep 27 0:00s should be removed from Europe/Minsk.
On 07/01/2016 11:03 PM, Stepan Golosunov wrote:
I suspect that lines 2:00 1:00 EEST 1991 Sep 29 2:00s 2:00 - EET 1992 Mar 29 0:00s 2:00 1:00 EEST 1992 Sep 27 0:00s should be removed from Europe/Minsk.
Yes, thanks, as you suspected those lines date back to when we had the wrong early-1990s transition times for Russia, and for Belarus as well. (I think they were from Shanks but do not have my copy of Shanks handy.) Evidently we fixed the former but not the latter. I installed the attached patch to fix this. This is all somewhat independent of the problem that started this thread, which was merely that the "0:00s" was confusing. Any time between 0:00 and 2:00 could have been put there, without changing the meaning of the table. Even the date could have been changed to July 1 without changing anything. It was a (confusing) no-change transition.
participants (4)
-
Ed Schouten -
Paul Eggert -
Stepan Golosunov -
Tim Parenti