FW: Wrong Argentinian DST introduced in version 2008f

[Aurelien Jarno]

On Mon, Oct 06, 2008 at 09:16:15PM +0700, Robert Elz wrote:
>     Date:        Mon, 6 Oct 2008 09:31:03 -0400
>     From:        "Olson, Arthur David (NIH/NCI) [E]" <olsona at dc37a.nci.nih.gov>
>     Message-ID:  <B410D30A78C6404C9DABEA31B54A2813029A0223 at nihcesmlbx10.nih.gov>
> 
>   | From: Aurelien Jarno [mailto:aurelien at aurel32.net]=20
>   | Sent: Monday, October 06, 2008 9:29
>   | To: tz at lecserver.nci.nih.gov
>   | Subject: Wrong Argentinian DST introduced in version 2008f
>   | 
>   | tzdata version 2008f introduced a rule for Argentinian DST based on
>   | http://www.worldtimezone.com/dst_news/dst_news_argentina03.html
>   | and
>   | http://www.impulsobaires.com.ar/nota.php?id=3D57832
> 
> Not really.  2008f changed the expected start date based upon those.
> Previous version(s) (certainly 2008e, maybe earlier I didn't check) had
> the start date from the first Sun in Oct, which is how the legislation
> apparently passed last December was explained (back last December.)

I have been told the legislation was for one year only, not for this
year.

>   | Unfortunately those links only speak about rumors of a DST change, there
>   | is nothing fixed officially, and according to some Argentinian people
>   | DST may even not happen this year.
> 
> Unfortunately, summer time is one of those things that people argue
> about a lot, and last minute changes happen all the time.  Sometimes
> there's no choice but to guess at what looks most likely, based upon
> all available information, as expecting to get updates distributed, and
> installed, everywhere, with just a couple of days (or even weeks) notice
> is absurd.

IMHO, it is a really bad idea to "guess" the next DST when we don't have
official information. People from Argentina really know the issue with
their government and they are ready to cope with last minute changes.

The problem with guessing is that the changes happens at a wrong date
and at a moment when it is *totally unexpected*. On the contrary without
trying to guess, the new version of tzdata may comes very late, but at
least people are aware of possible problems, and *expect* them.

> Which is exactly why there's all this trouble in Argentina now, if we
> had just had 2008f times, no-one would have noticed anything yesterday.
> It is just that 2008f isn't installed in enough places yet (it is only
> 3 weeks old) to have affected much - many sites are still using older
> data from when we believed that the first Sun in Oct was the right rule.

You can't ask people to be always up to date on the tzdata version.
People update tzdata when they now a DST change will happen, not when a
date that has been guessed and thus is *unknown* from the lambda user 
comes.

> Hence, some systems turned on summer time yesterday.  Others running
> either much older tzdata (last year) won't have summer time at all, and
> others running 2008f (or today's 2008g) will start it in 2 weeks.

So maybe in two weeks we will have the same problem again...

> A huge mess.   If there is anyone actually in Argentina reading this,
> (Aurelien I don't know where you are, nor Nicolas Alvarez)
> could you please contact the appropriate govt office and ask - ASAP.
> Get a reference to something official, written - if available via the
> web, even better, but even just an authoritative verbal statement would
> be better than nothing.
> 

I am not in Argentina, I am just relaying requests from Argentinian
people. Currently there is no official information available because *no
decision has been taken*. 

As usual, the decision will be taken very late regarding to the date of 
the change. What they want is that people *stop trying to guess* a 
possible date for the DST change, and instead just update tzdata when we
have official data, even if it is really late.

Regards,
Aurelien

-- 
  .''`.  Aurelien Jarno	            | GPG: 1024D/F1BCDB73
 : :' :  Debian developer           | Electrical Engineer
 `. `'   aurel32 at debian.org         | aurelien at aurel32.net
   `-    people.debian.org/~aurel32 | www.aurel32.net