Re: Index AM change proposals, redux

Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:

> * For Highly Non-Unique Data we can use Duplicate Compression
> The latter is the technique used by Bitmap Indexes. Efficient, but not
> useful for unique/nearly-unique data

I heard that GIN has already had duplicate-compression feature.

http://www.sai.msu.su/~megera/oddmuse/index.cgi/Gin
| Gin consists of a B-tree index constructed over entries (ET, entries tree),
| where each entry is an element of the indexed value (element of array,
| lexeme for tsvector) and where each tuple in a leaf page is either a
| pointer to a B-tree over item pointers (PT, posting tree), or a list of
| item pointers (PL, posting list) if the tuple is small enough.

If GIT comes, can we merge or share some modules between btree and gin?
I guess the page layout of GIT is better than ET/PT pair when the index
size are larger than main memory because the key and item pointers are
placed in near pages. Gin-over-btree might be useful some usages of
inverted indexes.

Regards,
---
ITAGAKI Takahiro
NTT Open Source Software Center