FW: About rules for Brazil

[Paul Eggert]

> From: Paulo Alexandre Pinto Pires [mailto:p at ppires.org]
> Sent: Monday, October 21, 2002 12:16 PM
> 
> Reading the source file for timezone data for Brazil is a joke.

It's an amusing story that has been repeated in many countries.
Here's a (non-Brazil) example that I plan to incorporate in my next
set of patches, along with a correction to America/Phoenix for 1944:

  Between 1944-01-01 and 1944-04-01 the State of Arizona used standard
  time, but by federal law railroads, airlines, bus lines, military
  personnel, and some engaged in interstate commerce continued to
  observe war (i.e., daylight saving) time.

So it's not just a story about the mayor of Rio de Janeiro whose name
I can't recall; it's also about Sidney P. Osborn, governor of Arizona
from 1941 to 1948, a powerful politician who fought the federal
imposition of daylight-saving time.


> I believe that if clocks eventually go wrong, it would be better if they
> just remained in the regular time.

In this month's case you would be right, of course; but I suspect that
it will be just as common for the clocks to be advanced ahead of our
guess as behind, and there your proposed strategy would fail just as
badly as the tz database's current approach.

So the question is: is it more common for DST changes to be
unexpectedly delayed (where your proposed strategy wins), or for DST
changes to come through at the guessed time (where the current tz
strategy wins)?  To answer this question, I compared the current
prediction for Brazil (Oct Sun>=8 to Feb Sun>=15) to the actual dates
since this prediction was added to the tz database on 1999-02-01.  We
have:

  actual date:  The predicted date was:
  1999-02-21    correct
  1999-10-03    late
  2000-02-27    early
  2000-10-08    correct
  2001-02-18    correct
  2001-10-14    correct
  2002-02-17    correct
  2002-11-03    early

So your proposed strategy would have been better on two occasions,
worse on five, and the same on one.

A similar situation obtains in Israel, where the guesses after 2004
are quite arbitrary.  However, for the few applications that need
access to future civial time (planning airplane trips?) I think it
better to have a good guess than to guess that there will be no DST at
all.

> Besides that, let those of us who are Brazilians try to push the
> goverment into using permanent criterion

You might look into Arizona's current solution: no DST whatsoever.