TZ changes for Brazil for 2003/2004

Paulo Alexandre Pinto Pires p at
Mon Sep 22 05:46:59 UTC 2003

Paul Eggert wrote:

>At Sat, 20 Sep 2003 20:18:48 +0200, Oscar van Vlijmen <ovv at> writes:
>>The information in
>>is a RECOMMENDATION from the Ministry of Mines & Energy.
>>Very likely a presidential decree will follow some time later.
>Is it plausible that any presidential decree will differ from
>I don't know Brazilian politics, unfortunately; nor can I read
>Portuguese.  But I don't see anything in that press release that
>indicates that it's merely a recommendation, and might change.
I couldn't get the Ministry's material (apparently they are too heavily 
loaded, down for maintenance or even under DoS attack), but the Ministry 
is the authority on subjects like energy savings, which is the purpose 
of summer time.  It would be surprising if the president would go 
against such recommendation, especially given energy shortage in recent 
years.  However, last year's last-minute change of summer time start 
date due to electoral authorities having forgot about summer time when 
programming hundred-thousands electronic poll devices is a sign that 
stupidity can always overcome logic and good practice (yet having dates 
for summer time change every year is not much less stupid).

>>To me it is not clear what the start-ending times are.
>>Is there any proof that 0 h wall clock time and 24 h wall clock time are
>>used for _all_ time zones?
>That's the natural assumption, as it's what seems to have been done in
>the past.  I'd expect more-specific language if they wanted all the
>clocks to change simultaneously.
If I understand it right, Oscar's doubt is if all time zones would 
change their clocks at 00:00 local time or 00:00 Brasília time.  Text 
for previous decrees seems pretty clear to me that local time is to be 
used, since the "legal time", the time used in legal affairs, is always 
that of the jurisdiction of each affair.  Also, Brazilian practice has 
always been change clocks in Saturday to Sunday midnight, to cause 
minimal impact on industrial and commercial activities.

>>See also decree no. 2784 from 1913 about the definition of the "legal hour".
>Sorry, I don't have access to that document.
<>.  That law establishes that:


      The Greenwich meridian is the fundamental meridian for contracts
      and commercial relations in Brazil.


      The country is to be divided in four distinct time zones (GMT-2 to
      GMT-5 for the different parts of national territory) for the
      purpose of legal use of time (to my undestranding, that means the
      contracts and relations above).


      Any previous law regarding the same subject is revoked.

There is no definition for "legal hour" as one specific clock time or 
anything like this.  I didn't see Oscar's point.

>>There are no DST related decrees after no. 4399 on the presidential decrees
>>lists for 2002 and 2003, up to decree no. 4842 from Sept. 18 2003.
>>I'd suggest: keep watching the latter page!
>That's a good suggestion.  But should we hold up on updating the tz
>database until after a decree is issued?
As a national of Brazil, what I can say is that summer time start and 
end dates, as well as where it will be used, have already been widely 
announced by the press.  It would be very annoying for everyone if the 
government would go back and change anything now (but it has already 
happened at least twice in the past, so I'd not be surprised).

Perhaps I am too passionate, but this year there is one more problem 
that affects users, besides changes in dates: changes in zones 
themselves.  Peoples with systems in Bahia or Mato Grosso do Sul with 
have to stop linking their /etc/localtime to 
/whatever/zoneinfo/Sao_Paulo or /whatever/zoneinfo/Cuiaba, respectively. 
 The sooner tzdata is updated the more likely it is that changes be 
deployed in time, and perhaps even included in new releases of the 
different operating systems that uses tz.  In 2002, summer time in 
Brazil started only on 2002/11/03, but prevous tzdata releases had a 
rule that caused summer time to start on 2002/10/13, and an updated 
tzdata was only released on the 2002/10/15.  For two days, official 
tzdata was causing clocks to be wrong.


... Qui habet aurem audiat quid Spiritus dicat ecclesiis.

More information about the tz mailing list