I just noticed that from a C or C++ program using
libpq or libpq++, I can send *one* command that
contains several SQL statements separated by
semicolon. Something like:
PgDatabase db (" .... ");
const char * const sql =
"insert into blah (...); insert into blah (...)";
if (db.Exec (sql) == PGRES_COMMAND_OK)
{
cout << "Fine!" << endl;
}
And I verify the table, and all the inserts took place
(and of course, the program outputs "Fine!").
But I'm wondering -- is this a PostgreSQL extension,
or is it "legal SQL"? In particular, I'm wondering
if it is a feature that in the future you might
decide to eliminate for not being ANSI-SQL compliant.
What happens if the first command is ok but the second
one fails? I guess PgDatabase::Exec would return an
error code, and PgDatabase::ErrorMessage would return
the error message corresponding to the second statement
(the one that failed). Am I correct in thinking this?
Any reason why this should be avoided? (on the plus
side, I think this might increase efficiency for
transactions where one executes several insert or
update statements).
Thanks for any comments,
Carlos
--
---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster
