proposed time zone package changes (Argentina, Mauritius, Niue, Syria)

Robert Elz kre at munnari.OZ.AU
Thu Oct 9 12:48:36 UTC 2008


    Date:        Wed, 08 Oct 2008 13:22:49 -0300
    From:        Nicolas Alvarez <nicolas.alvarez at gmail.com>
    Message-ID:  <gcimog$sa8$1 at ger.gmane.org>

  | So basically the current tzdata has a *guess* at when it *might* be.

Everything in tzdata that claims to indicate what will happen in the
future is a guess as to what might happen.   How can it be otherwise?

Unless the timezone code says nothing at all about times from the date
of its creation (that is from the reeast time of each tzdata update - so
the "date" command on your system would always give the current time in UTC)
it cannot really do otherwise.   None of is is able to predict the future
with certainty, all we can ever do is guess.

Now some of the guesses are ones we can be reasonably confident about, like
when summer time will end in the US this year, others are  total speculation
as we get no data at all on what is happening, and many others are
somewhere in between - but they are all guesses, and if we were required to
avoid using any guesses (predictions) at all the whole thing would be a
disaster.

  | As you probably know, many operating systems had their time wrong last week
  | because tzdata said DST would start on the 5th. Ubuntu took days to update
  | the tzdata since the report. The current package says it will start on the
  | 19th.

Yes, we try hard to get the best information we can, and make the most
reliable estimations of what is likely to happen, and most of the time
things work out the way we'd like.   Sometimes they don't.  What really
matters is the accuracy of the information that we can obtain, and how
early that information reaches us.   We know there are places where
summer time rules seem to be decided upon a whim the day before the event,
but I suspect that most of this is just perception - things like airline
timetables for international flights cannot simply be changed with a
day's notice.   So what is most likely the problem is just that the information
is not getting through, rather than it does not exist.

That is, I am reasonably confident that there are people in Argentina
who really know whether summer time will start Oct 19 this year or not,
and if not, when (if at all.)   The issue is getting that information from
the people who know and to the people who maintain tzdata.   Someone who
cares needs to do that, this year, and in the future (until perhaps things
stabilise and there are less surprises.)

  | If the government doesn't say *anything* about DST,

Why would they not?   That is, if asked?   If no-one bothers to ask,
that's a different issue, that is is you just sit passively and hope to
be informed, but why not go out and ask???

That's what I asked for in an earlier message - no-one seems to have
done anything (beyond speculating.)

  | what will you do? Change the rules on the 18th to remove the DST?

No, because on the 18th we still will not know one way or the other, on
this assumption (that no-one bothers to find out in advance.)   What would
happen is that on the 19th we will find out what actually happened, and if
we were wrong, we will correct it then.

That's obviously not ideal, but if no-one is going to say any more than
"I haven't heard anything yet" then what choice is there?   The reason you
aren't hearing anything may be because they assume you already know from
earlier announcements (the actual "prepare to change your clocks" messages
usually only start appearing a day or two before the actual event, even in
places with very stable summer time rules.)

Note here that no-one simply decided that Oct 5, and then Oct 19,
would be good days to screw with the Argentinian people.  They came
from the most reliable information we have been able to find so far.

The legislation that said summer time would start in Oct was last
December, so it cannot reasonably have meant Oct 2007, it must have
meant 2008.   It has been suggested that that legislation has been
repealed, if so, find some authoritative source - ie: get someone who
really knows to say something (and ideally a reference to some official
doc, if it is available on the net, even better) and send that.   Then
we will have more up to date, and reliable information, and the rules
will get updated (well, actually that's a guess too, as that's also in
the future, but it has always happened in the past...)

  | Do you know how much time it will take for all distros to have it?
  | Definitely more than a day, so on the 19th
  | we'll have the same mess again.

Yes.   Unfortunately.   Unless someone actually does the work to get
better information that's what is going to happen (assuming summer time 
does not start on the 19th, at the minute that seems more likely than
other scenarios).

kre




More information about the tz mailing list