Thanx Wayne and Pavel,
i will try to make this work with cursors. Theoretically there can be more
than 4 bilion records in result, but only 58mil returned which took 3 days.
Possibly i will have to create temporary table and store results form
%patterns into it after each 500k records, and group
I have found cursors example in plperl. now it works fine. the code is
below.
CREATE OR REPLACE FUNCTION uniprot_frekvencije.pattern_counter2(patLength
integer)
RETURNS character varying AS
$BODY$
my $sth = spi_query(select sequence from entry);
my $patLen = $_[0];
my $patt = '';