Oleg Bartunov wrote:
> Is't late to submit new contrib module for 7.2 ?
> It's almost ready and tested, we have to write README.tree.
> New module creates new data types 'tree', 'treequery' for tree-like data and
> provides indexed access methods using Btree or GiST.
> Node represents as a path to the root, so record like '3.4' means
> 4-th child of 3-rd child of the root node.
> Due-to bit-representation there is a limitation to the number of
> children - 64 and defined in compile time (valid values - 8,16,32,64).
How hard it would be to automatically expand it so that 65th node will
make it flow over to next bitfield and the whole level will then
accommodate 64*64 = 4096 child nodes ?
> For practical purposes 64 children looks quite enough.
> insert,delete operations should be ok, while move operations
> is a pain - one should update all paths of children nodes.
> But search is very fast, we use data from Open Directory catalog (www.dmoz.org)
> for testing:
Could you send it to me for personal testing if it is not accepted to
contrib for 7.2 ?
In response to
pgsql-hackers by date
|Next:||From: Tom Lane||Date: 2002-01-25 16:04:12|
|Subject: Re: New system OIDS inside include/catalog/pg_proc.h |
|Previous:||From: Daniel Kalchev||Date: 2002-01-25 15:58:28|
|Subject: Re: RTREE Index on primary key generated by a sequence |