Re: tsearch in core patch, for inclusion

>
> >> CREATE TABLE foo (id serial, names text FULLTEXT);
> >>
> >> Anything more complicated is a waste of cycles.
> >>
> >> Joshua D. Drake
> >
> > I agree. Question: what about multilanguage fulltext.
> >
> > CREATE INDEX foo USING FULLTEXT(bar) [ WITH czech_dictionary ];
> > CREATE TABLE foo (id serial, names text FULLTEXT [ (czech_dictionary,
> > english_dictionary) ] );
> >
> > all others can we do via SP.
>
>That works for me with perhaps a default mapping to locales? For example
>if our locale is en_us.UTF8 we are pretty assured that we are using
>english.
>

90% yes. 10% no. In czech typical task: find word without accents, or find
german, english, czech stemmed word in multilanguage documents (or different
languages depend on topology). Lot of databases are minimal bilanguagal (in
czech rep. german and czech).

Pavel

p.s. missing collates is big minus for PostgreSQL in eu (we have some
workarounds)

_________________________________________________________________
Najdete si svou lasku a nove pratele na Match.com. http://www.msn.cz/