Tom Lane wrote:
> Bruce Momjian <bruce@momjian.us> writes:
>> Tom Lane wrote:
>>> Since I don't think that a datatype solution is the way to go,
>>> I don't feel that we are as far away from an agreement as Bruce
>>> is worried about.
>
>> Well, from where I sit, there is one person saying give me the foot gun,
>> and Heikki saying he wants a bullet-proof type system, and you and I are
>> in the middle, so the big problem is I don't see a concensus forming,
>> and we have been discussing this for a while.
>
> The people who actually use tsearch2 seem to all have the same opinion ...
> so I think we can't go too far in the bullet-proofing direction.
>
> But I would like a design that is bulletproof in dump/reload scenarios,
> and I think it's fair to question that aspect of the tsearch2 design
> because we've seen many reports of people having trouble updating
> databases that use tsearch2.
dump/reload is *the* biggest problem I've had with tsearch2 so far. But
it hasn't been with the actual data - it's been the functions, and only
when migrating between versions. But solving dump/reload reliably is one
of the main things I'm hoping for in 8.3 ;-)
As for a nother use-pointer, I use different configurations in the same
database - but only one per table. I explicitly use the to_tsvector that
specifies a configuration always - to avoid surprising myself.
I don't use the functional index part, but for new users I can see how
that's certainly a *lot* easier. Requiring the specification of the
configuration explicitly when creating this index I don't see as a big
problem at all - compared to the work needed to set up triggers. But
it's nice not to have to do it when querying. But wouldn't that be
solved by having to_tsvector() require the configuration, but
to_tsquery() and plainto_tsquery() not require it?
//Magnus