On Thu, Jan 8, 2015 at 12:31 AM, Alexander Korotkov <aekorotkov@gmail.com> wrote:
Rewritten version of patch is attached. I made following changes:
1) I removed fillfactor handling from entry tree. Because in this case fillfactor parameter would be only upper bound for actual page fullness. It's very like GiST approach to fillfactor. But I would rather remove fillfactor from GiST than spread such approach to other AMs.
2) Fillfactor handling for posting trees is fixed.
3) Default fillfactor for GIN is reverted to 90. I didn't mean that default fillfactor should be 75%. I mean that equilibrium state of tree would be about 75% fullness. But that doesn't mean that we don't want indexes to be better packed just after creation. Anyway I didn't see reason why default fillfactor for GIN btree should differs from default fillfactor of regular btree.
Just forgot simple case that I used to test the patch.
create table test as (select array[1,2,3] v from generate_series(1,1000000));
create index test_idx20 on test using gin(v) with (fillfactor=20);
create index test_idx30 on test using gin(v) with (fillfactor=30);
create index test_idx40 on test using gin(v) with (fillfactor=40);
create index test_idx50 on test using gin(v) with (fillfactor=50);
create index test_idx60 on test using gin(v) with (fillfactor=60);
create index test_idx70 on test using gin(v) with (fillfactor=70);
create index test_idx80 on test using gin(v) with (fillfactor=80);
create index test_idx90 on test using gin(v) with (fillfactor=90);
create index test_idx100 on test using gin(v) with (fillfactor=100);
List of relations
Schema | Name | Type | Owner | Table | Size | Description