Re: [GENERAL] DECLARE CURSOR

snpe <snpe(at)snpe(dot)co(dot)yu> writes:
> When I call DECLARE CURSOR out of transaction command success,
> but cursor is not created
> Reference manual say that this get error :
> ERROR: DECLARE CURSOR may only be used in begin/end transaction blocks

Oops. I removed that test on 21-Oct as part of this fix:

2002-10-21 18:06 tgl

* src/: backend/access/transam/xact.c, backend/catalog/heap.c,
backend/catalog/index.c, backend/commands/dbcommands.c,
backend/commands/indexcmds.c, backend/commands/tablecmds.c,
backend/commands/vacuum.c, backend/parser/analyze.c,
include/access/xact.h: Fix places that were using
IsTransactionBlock() as an (inadequate) check that they'd get to
commit immediately on finishing. There's now a centralized routine
PreventTransactionChain() that implements the necessary tests.

My reasons for removing it were (a) it was in the wrong place (analyze.c
is not the right place to test execution-time constraints), and (b) it
was the wrong test: the test as written was just IsTransactionBlock(),
which is wrong in the case of autocommit-off, since a DECLARE CURSOR
will start a new transaction perfectly well. Another objection is that
inside a function call, it ought to be legal to do DECLARE CURSOR even
if we're not in a transaction block, since the function might intend to
use the cursor itself before returning.

I think I had intended to put together an alternative test that only
complained about interactive DECLARE CURSOR and understood about
autocommit, but I forgot.

At this point we can either add the fixed-up error check (meaning RC1
won't be the release after all), or change the documentation.

Comments?

regards, tom lane