20 April 2005 4 comments Linux
I just learnt a better way to check if a PostgreSQL table exists. Before I did this to find out if table mytable is defined:
SELECT * FROM 'mytable' LIMIT 1;
But this had to be wrapped in exception catching application code because I was able to tell if the table existed if the select statement worked. That's the wrong and naive way. The correct way is to use pg_class and look at how many rows were returned:
SELECT relname FROM pg_class
WHERE relname = 'mytable';
Why have I missed this before? No explanation?
A disadvantage with using pg_class is that it's not ISO standard, meaning that it's specific to PostgreSQL only. The first pattern will work on all relational databases.
SELECT 1 FROM pg_catalog.pg_class WHERE relkind = 'r' AND relname = 'name' AND pg_catalog.pg_table_is_visible(oid) LIMIT 1