Re: static genericcostestimate

Tom Lane wrote:

>"Ramy M. Hassan" <rhassan@cs.purdue.edu> writes:
>  
>
>>The genericcostestimate function is currently static. This limits the 
>>development of new access methods as loadable modules without touching 
>>pgsql sources. Currently I have to include a copy of the function in the 
>>module, which is obviously too bad.
>>Is there any reason to keep this function static ?
>>    
>>
>
>Is it really of much use for your access method?  It's such a crude hack
>that I didn't want to encourage people to use it ... it is really just a
>stopgap until someone gets around to thinking harder about the actual
>access behavior of the existing index AMs.
>
>BTW, what are you working on?  I had no idea that anyone was
>experimenting with new index methods.
>
>  
>
I am currently working on porting SP-GiST to postgresql.
SP-GiST is an adaptation of GiST to support space partitioning trees ( 
http://www.cs.purdue.edu/homes/aref/dbsystems_files/SP-GiST/ )
The current standalone SP-GiST implementation is based on libgist v1.0 
from berkeley ( http://gist.cs.berkeley.edu/libgistv1/ )
The core SP-GiST is being implemented as module to be loaded before any 
spgist extention module.
I am expecting the first alpha release early of May.
Currently, there is no effort done in cost estimation for SP-GiST, so 
the genericcostestimate seams to be ok for now.

>            regards, tom lane
>  
>