A problem here is that _create_hash_desc is called many times to
create a TupleDesc with int32 attribute.
I've tried to implement the function like this ,
TupleDesc _create_hash_desc()
{
static bool firstcall = true;
static TupleDesc tupdesc;
if(firstcall){
tupdesc = CreateTempl
I've fixed the patch just now. It works and pass the regression test ;-)
Here is the new patch. I'll keep the hash code in order and use
binary search in a later version soon.
diff --git a/src/backend/access/hash/hash.c b/src/backend/access/hash/hash.c
index 6a5c000..1a8dc75 100644
--- a/src/bac
Hi, hackers.
I'm working on my gsoc project of improving hash index.
I posted a thread to ask some questions about implementation a few hours ago.
http://archives.postgresql.org/pgsql-hackers/2008-07/msg00721.php
Someone's advice reminds me of the problem of design decision again.
I store hash in
