Tom Lane wrote:
> Alvaro Herrera <alvherre@commandprompt.com> writes:
> > Bruce Momjian escribi�:
> >> What has really hurt the default GUC idea is how to do restores from a
> >> pg_dump.
>
> > I guess what should happen is that pg_dump should include a
> > SET default_text_search_config = 'foo'
> > just before the CREATE INDEX, like we do for other variables.
>
> The nasty cases are in data-only dumps, that is, where you're trying to
> load data into a table with pre-existing indexes or triggers. A SET
> like the above is at least as likely to be wrong as right, if the index
> or trigger depends on it to tell it what to do.
Ouch. I had not even thought that far.
FYI, yes, the default tsearch GUC controls operations per row _if_ you
have triggers or expression indexes that rely on the default
configuration GUC. If you have specified the configuration, there is no
problem, and hence my conclusion that the default GUC is too
error-prone.
-- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB
http://www.enterprisedb.com
+ If your life is a hard drive, Christ can be your backup. +