Обсуждение: tsearch2 's access method?

the basic structures to implement  the index is :

1.B- tree or the alike.

2.hash structures

So the question is arise:In the tsearch2 model, I thought the general access method is GiST tree, a tree this like B- tree.

 But when I saw the crc32_sz() function to hash the tsvector type, the index implementation again turns to be a hash list,

which is the right? hash structures or tree like B tree to implment the index? 

沈一枫 wrote:
> the basic structures to implement  the index is :
> 
> 1.B- tree or the alike.
> 
> 2.hash structures
> 
Not only. R-tree, SS-tree, kd-tree, RD-tree etc


> So the question is arise:In the tsearch2 model, I thought the general 
> access method is GiST tree, a tree this like B- tree.
> 
>  But when I saw the crc32_sz() function to hash the tsvector type, the 
> index implementation again turns to be a hash list,
> 
>
> which is the right? hash structures or tree like B tree to implment the 
> index? 

None of them. Tsearch2 use signature tree, which is very close to RD-tree.


GiST (its realisation in pgsql) can store tree likes to R-tree or B-tree (with 
some limitations, such as now pgsql's GiST doesn't support order).

-- 
Teodor Sigaev                                   E-mail: teodor@sigaev.ru
  WWW: http://www.sigaev.ru/