On Wed, 12 Sep 2001, Matt wrote:
> create table tez (i int);
> insert into tez values(3);
>
> create function tezt0(int) returns int as '
> begin work;
> lock table tez;
> select * from tez as r;
> commit work;
>
> select tez.i;
> ' language 'sql';
It looks to me from testing reasonably recent sources
that the above really isn't safe. Since we don't have nested
transactions, that's not a separate transaction from whatever
it's running in, but instead the begin will NOTICE if you're in a
transaction and the commit will commit it and now you're no longer
in a transaction...
begin;
select tezt0(4);
insert into tez values (4);
rollback;
will end up with tez getting the row and a notice about the fact
you weren't in a transaction from rollback. Is this really reasonable
behavior?
---------------------------(end of broadcast)---------------------------
TIP 3: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to [EMAIL PROTECTED] so that your
message can get through to the mailing list cleanly
