Subject | Re: [Firebird-Architect] Re: RFC: Please unify stored procedure execution |
---|---|
Author | Dmitry Yemanov |
Post date | 2004-12-22T07:51:46Z |
"Dimitry Sibiryakov" <SD@...> wrote:
2) It's described depending on its output params existence to be either
select or execute (like EXEC BLOCK behaves)
3) You either execute it immediately or fetch rows + do DSQL_close depending
on (2)
Dmitry
>Nope. We'll need a new iteration construct for PSQL.
> >4) Add CALL statement (BTW, this is a standard syntax for procedures,
> >see SQL-99) which cares about output parameters and return either a
> >single set of outputs or a cursor and hence will be described as
> >either select or execute. But what should be a PSQL syntax to retrieve
> >a cursor? Or will this be a client-only thingy?
>
> If I'm not mistaken, you implemented full cursors' support for PSQL
> in 2.0 including DECLARE CURSOR, FETCH and so on. May CALL return a
> CURSOR?
> What should return the procedure above if it is called with CALL?1) CALL can be used for all types of procedures
> Should a coder use isc_dsql_free_statement() with DSQL_close after
> executing CALL statement? Should data in buffers be filled right
> after call isc_dsql_execute() or isc_dsql_fetch() is required?
2) It's described depending on its output params existence to be either
select or execute (like EXEC BLOCK behaves)
3) You either execute it immediately or fetch rows + do DSQL_close depending
on (2)
Dmitry