On Jul 26, 2009, at 1:32 PM, Ivan Sergio Borgonovo wrote:
Actually a serializable transaction doesn't even seem a too bad
solution... but I just would like to understand better how to manage
this situation so that I could make it as simple as possible AND
lower as much as possible the chances tha
Suppose I've
create table m1 (
a int primary key,
cm1 int,
cm2 int,
...
);
create r1 (
a int references m1 (a),
cr1 int,
cr2 int,
...
);
and cm1 possible values depends on some function of cr1 for a given
a.
I actually have a plpgsql function that returns the possible choices
for cm1 fo
