Re: About "Our CLUSTER implementation is pessimal" patch

> If we ever get another index type that supports ordered
> scans, it'll be time enough to worry about cases like this.


Ok

> BTW, I think you could use tuplesort_begin_index_btree() rather than
> touching _bt_mkscankey_nodata directly.


well I created my own tuplesort_begin_rawheap method (copied from:
http://archives.postgresql.org/pgsql-hackers/2008-08/msg01371.php ).
From there I call _bt_mkscankey_nodata (as tuplesort_begin_index_btree()
does), plus I set up everything else I'm going to need in tuplesort.

Another question:

why is IndexInfo.ii_Expressions a list? How can an index have more than
one expression? Sorry if it's a stupid question, but I'm not familiar with
index expressions.

I think I'm almost there (some very stupid tests pass). I'll try to submit a
patch soon to understand if I'm going in the right direction.



Leonardo