Re: Second-granular timezone offset format not documented

On Mon, Jul 5, 2021 at 9:56 AM PG Doc comments form
<noreply@postgresql.org> wrote:
> Note how the response has a very weird timezone offset. I guess it is valid,

As for whether it's valid, that's coming from the IANA tz dataset.  It
has a moment that it believes standard time to have begun at each
location, in this case:

Z America/Mexico_City -6:36:36 - LMT 1922 Ja 1 0:23:24

https://en.wikipedia.org/wiki/Time_in_Mexico#History seems to agree on
the year at least.  That "local mean time" offset is computed from the
location's longitude, for lack of anything better.  The tzinfo
"Theory" file has a bunch of disclaimers about pre-1970 data though,
including "the tz database's LMT offsets should not be considered
meaningful".