On 7/5/19 5:39 AM, Andrey Sychev wrote:
> Thank you very much for answering my question, Tom.
> Yes, I have always assumed that returning from function without
> calling SPI_freetuptable is not good idea, but I do not know another
> way to achieve same result.
Please do not top post on the Postgres lists.
> 1. As I understand there are tests for SFRM_Materialize in code above.
> In my case a caller of my function is PL/pgSQL procedure.
> Does it accept this returning mode?
See contrib/tablefunc/tablefunc.c crosstab_hash() as an example.
> 2. Our current production server versions is 9.1 and 9.6.
> Do this versions support returning of tuplestore?
Works since Postgres 7.3 if I recall correctly.
> 3. Currently my function defined as "RETURNS SETOF".
> Does definition of the function need to be changed if I rewrite code to
> return tuplestore?
No
HTH,
Joe
--
Crunchy Data - http://crunchydata.com
PostgreSQL Support for Secure Enterprises
Consulting, Training, & Open Source Development
