Re: Trigram (pg_trgm) GIN index not used
От | Merlin Moncure |
---|---|
Тема | Re: Trigram (pg_trgm) GIN index not used |
Дата | |
Msg-id | CAHyXU0xmLCTQpYkD8uZhw2=T8JUqcfucO2Q+9B4bGiJZr+hLAg@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: Trigram (pg_trgm) GIN index not used (Tom Lane <tgl@sss.pgh.pa.us>) |
Ответы |
Re: Trigram (pg_trgm) GIN index not used
|
Список | pgsql-general |
On Thu, Feb 21, 2013 at 9:57 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote: > Ivan Voras <ivoras@freebsd.org> writes: >> I have a table with the following structure: >> ... >> raw_data | citext | not null >> ... >> "documents_raw_data_trgm" gin (raw_data gin_trgm_ops) > >> I'd like to use pg_trgm for matching substrings case-insensitively, but >> it doesn't seem to use the index: > > You're outsmarting yourself by using citext as the column datatype. > That causes "ilike" to be interpreted as a citext-specific operator, > which is not a member of the gin_trgm_ops operator class, so it doesn't > match this index. > > I wonder whether we really need that citext-specific operator at all > ... but in the meantime, if you need the column to be citext for some > other reason, I'd suggest making a gin index on raw_data::text and > then writing the query as raw_data::text ilike '%zagreb%'. hm, one more data point that citext implementation didn't succeed in terms of abstracting you from case sensitivity issues. merlin
В списке pgsql-general по дате отправления: