Iñaki Baz Castillo wrote:
2009/9/16 Alex Balashov abalashov@evaristesys.com:
Iñaki Baz Castillo wrote:
For example a PostgreSQL BIGINT is unintelligible to sqlops, I found out.
What do you mean exactly?
CREATE TABLE stuff ( id bigint PRIMARY KEY NOT NULL );
INSERT INTO stuff VALUES (1);
...
sql_query("ca", "SELECT id FROM stuff", "ra");
$dbr(ra=>[0,0]) will be NULL.
sql_...ops!
A bug?
Maybe. There was mailing list traffic about it when I first reported it around 1 April 2009.
I do not think it is a bug, but rather the understandable unwillingness of the implementors to handle every database-specific data type that might come their way, like 'timestamp', 'inet', etc.
I do think 'bigint' is an important one to handle because there are some values Kamailio may need to retrieve from time to time that are simply larger than 32-bit width. But it's hard to say where the line should be drawn on special RDBM types, especially since the values that are represented by user variables, AVPs, hash table, etc. are ultimately just scalar quantities or strings.