Hello
this patch significantly reduce a ccache searching. On my test - bubble sort
postgres=# \sf buble
CREATE OR REPLACE FUNCTION public.buble(integer[])
RETURNS integer[]
LANGUAGE plpgsql
AS $function$
declare
unsorted bool := true;
aux int;
begin
while unsorted
loop
unsorted := false;
for i in array_lower($1,1) .. array_upper($1,1) - 1
loop
if $1[i] > $1[i+1] then
aux := $1[i];
$1[i] := $1[i+1]; $1[i+1] := aux;
unsorted := true;
end if;
end loop;
end loop;
return $1;
end;
$function$ immutable
it decrease evaluation time about 15%.
Regards
Pavel Stehule
p.s. I know so bubble sort is not effective for large arrays. This
algorithm was used because a array is intensive modified.