[...]
> This makes sense to me, yes. The reason why I'm a bit lost is that
> we never did anything whatsoever with respect to encoding. Oddly
> enough, I couldn't find much about this in the docs.
> [...]
>
> I guess what changed from version 7.4.x to 8.0 is that the default
> server_encoding changed?
Some further digging through the documentation reveals (in addition
to the fact that I seem to be an incompetent when it comes to search
through documentation :-)) that the above is indeed the case.
Each database apparently has its own encoding, that can be specified
with either the createdb shell command (switch -e or --encoding, IIRC),
or with the SQL command create database.
With version 7.4.x, when I execute the command:
show server_encoding;
I get SQL_ASCII as response. Also, when I do:
select * from pg_database;
I get a bunch of columns where the encoding field contains 0 for all
of them (I never specified encoding when creating databases, so in
all those cases it took the default).
With 8.0.1, the command show server_encoding returns UNICODE as
result -- this is also the default, since I created the database
without explicitly indicating encoding.
Apparently, conversion from latin1 to UNICODE works? (or at least
doesn't trigger an error -- I wonder if some characters appear now
incorrectly? I'll check that).
Yes, I know I'm blah-blah-ing and answering my own questions :-)
I'm just posting the progress I've made in case someone else was
also interested in finding out more about the problem or possible
solutions.
Thanks,
Carlos
--