We found the problem and will submit a patch for 7.2.1.
We were confused by 'index toasting' and thought index is realy toasted
but actually it's only compressed. So, we have to rollback our
changes for intarray and tsearch :-( We were happy to make gist index
non-lossy, but it's not the case for long data. Probably it'd be
possible to define in compile time if index will be lossy.
Tom, does index compressing is a temporal solution ? Do you intend
to implement real toasting ?
Oleg
On Mon, 4 Feb 2002, Tom Lane wrote:
> Oleg Bartunov <oleg@sai.msu.su> writes:
> > we've got a message from Poul about problem with our contrib modules
> > tsearch and I dont' know yet if this could be really tsearch problem,
> > it's time to go to the bed :-) Could be this problem related to the
> > partial index he used ?
>
> Seems unlikely. But in any case, you'd be more likely to get help
> by giving a complete script for reproducing the problem. I'm not
> feeling like reverse-engineering Poul's table schema and then guessing
> at his data too...
>
> regards, tom lane
>
> ---------------------------(end of broadcast)---------------------------
> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org
>
Regards, Oleg
_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83