On Thu, 18 Nov 2004 11:08:38 +0100, Markus Wollny
<markus.wollny@computec.de> wrote:
> Oleg, what exactly do you mean by "tsearch2 doesn't support unicode yet"?
> It does seem to work fine in my database, it seems:
> ./pg_controldata [mycluster] gives me
> pg_control version number: 72
> [...]
> LC_COLLATE: de_DE.UTF-8
> LC_CTYPE: de_DE.UTF-8
Correct me if I am wrong, but I think that UTF-8 is almost identical
to ISO-8859-1 in binary form to ISO-8859-1. I mean, UTF-8 is
ISO-8859-1 plus multibyte characters from other charsets.
If I am correct, there is no difference for Tsearch2 between UTF-8
and ISO-8859-1, so German locales work fine. As for ISO-8859-2
and similar this will not be the case since there are multibyte
characters.
Oh, and a side-question -- is there any facility to "strip" charsets
like ISO-8859-2 down to ASCII (not SQL_ASCII, I mean ASCII, only
latin characters without accents, etc.)? I think it would be useful
for me, since I intend to search through mail-like messages and
many people are lazy (or accustomed to...) and don't use special
chars but only their ASCII non-accented counterparts...
Regards,
Dawid