[Broken SQL instead of performance issue fixed]
It would appear that when I define the index on the FTI table (string and
oid) to be unique (which makes sense, since there is little point in having
duplicate rows in this case), a lot of inserts fail where they shouldn't. I
am guessing that if
Gordan Bobic <[EMAIL PROTECTED]> writes:
> [ why is this slow? ]
> explain select jobs.title from jobs, jobs_description_fti,
> jobs_title_fti where (jobs_description_fti.string = 'linux' or
> jobs_title_fti.string = 'linux') and (jobs_description_fti.id = jobs.oid and
> jobs_title_fti.id = job
