tzdata2006n Brazil incorrect fall-back date 2009-mar-01

[Paul Eggert]

"Dave O'Callaghan" <daveoc at uk.ibm.com> writes:

> I have been alerted to this website 
> http://icu-project.org/trac/ticket/5470 which describes an issue with the 
> Brazil update in tzdata2006n. 
>
> Is this true?

It appears to me to be a problem in the ICU4C's code simpletz.cpp, not
in the tz code or data itself, so you should ask whoever wrote
simpletz.cpp to fix it.  The tz data currently says that the fall-back
transition in Brazil in 2009 is the last Sunday in February, which
will be 2009-02-22, but the ICU4C (?) code apparently does not handle
this case correctly.

To some extent this is all academic.  Judging by past experience,
Brazil most likely will change the rules before 2009 rolls around, so
the tz data is just guessing about the 2009 transition.