Dear Tom:
Sorry, but I think you can see the problem without the definition of the
function. If after this message you continue thinking you need the source of
the function I will send it to the list. I don't want to extend so much the
message.
There is a master table (tablea) and a detail table (childtablea). With
"otherfunction" I want to retrieve in a string all the values related with a
row in the master table, separated by a "-".
select otherfunction('childtablea', 'fielda', fielda) from tablea;
So, otherfunction takes three parameters, the first is the name of a detail
table (childtablea), the second is the name of a foreign key in that table
(fielda) which references a primary key (also called fielda) in a master
table (tablea), and the third is the value of the foreign key I'm looking
for.
"immutablefunction_a" and "immutablefunction_b" are both wrappers functions
for "otherfunction", because I can't create a functional index over a
function with static arguments.... (This is true, isn't it?????)
So, the definition of immutablefunction_a is:
create function immutablefunction_a(int) returns text as '
begin
return otherfunction('childtablea','fielda', $1);
end;' language plpgsql immutable;
create function immutablefunction_b(int) returns text as '
begin
return otherfunction('childtableb','fieldb', $1);
end;' language plpgsql immutable;
So, the "bug" is that the when the result for immutablefunction_a(10) is,
for example, "10-20-30" and the result for immutablefunction_b(30) is
"10-20-30", and if this is used in the join clause, records are not
joined....
I hope you can understand me, I know is a little complicated schema....
Thank you very much.
P.D.: Tom, is the first time I speak to you, and I want to express my
congratulations for the work you and your folks are making. I'm very pleased
with the product.
-----Mensaje original-----
De: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Enviado el: jueves, 22 de junio de 2006 16:55
Para: Pedro J. Romero
CC: pgsql-bugs@postgresql.org
Asunto: Re: [BUGS] BUG #2487: Immutable functions results
"Pedro J. Romero" <multiacademia@eresmas.com> writes:
> Sorry if this not a bug, but I think so:
This report is useless, since you have not shown us either function.
regards, tom lane