I have been working with the 2000h version of the time zone and daylight
savings time data at
ftp://elsie.nci.nih.gov/pub/tzdata2000h.tar.gz , and
ftp://elsie.nci.nih.gov/pub/tzcode2000h.tar.gz .

I have found some unusual results in the rules files for time zones
in Brazil.  I have given a high level statement of the unusual, and
then detailed output from the zdump command later.
The details are in the attachment.

Do you get similiar results from the zdump command?
Have I made a mistake in understanding the zdump results?
Is there any way that I can help clear this up?

***  Local Name BRT 3 BRST in Brazil
The time zone America/Recife has a seven day long Daylight
Savings Time (DST) in October 2000, and no DST in 1991-1998, 2001,
and beyond.

The time zone America/Maceio has a fourteen day long DST
in October 2000, and no DST in 1991-1994, 1997, 1998, 2001,
and beyond.

The time zone America/Fortaleza has a fourteen day long (DST)
in October 2000, and no DST in 1991-1998, 2001, and beyond.

The time zones America/Araguaina and America/Sao_Paulo have
the same local names as the three time zones above.  They are all
in Brazil.  These time zones have reasonable length DST in 2000,
and 2001.  These look OK.

***  Local Name AMT 4 AMST in Brazil
The time zone America/Boa_Vista has a seven day long  (DST)
in October 2000, and no DST in 1988-1998, 2001, and beyond.

The time zone America/Cuiaba has the same local name.  They
are both in Brazil.  This time zone has reasonable length
DST in 2000, and 2001.  This looks OK.

-- 

Regards, Hank Ivy

| Henry (Hank) B. Ivy           PHONE : (916)785-1392             |
|                               FAX   : (916)785-9238             |
| ecardfile : https://ecardfile.com/    id : hank_ivy             |
| Hewlett-Packard Company       USENET: hivy@rosemail.rose.hp.com |
| 8000 Foothills Boulevard, MS 5617                               |
| Roseville, California 95747-5617                                |