On 1/9/2011 5:38 PM, chris.waldmann wrote:

>
>
> --- In Firebird-Architect@yahoogroups.com, "tempo93fr"<makowski@...> wrote:
>> there are also other possibilities , not only R*Tree
>>
>> for example Nested Containment List (NCList)
>> : http://bioinformatics.oxfordjournals.org/content/23/11/1386.abstract
>>
>> but also as Mongodb (http://www.mongodb.org/display/DOCS/Geospatial+Indexing), using geohash (http://en.wikipedia.org/wiki/Geohash
> It looks like there are many useful indexing kinds available, like Hilbert R*Tree, NCList,...
> http://www.cise.ufl.edu/~jgvenkat/files/MAMSurvey_jay.pdf
>
> So I think the great solution would be the support of GiST for alternative (and B-Tree?) indexing. GiST encapsulates core database indexing functionality including search, update, concurrency
> control and recovery.
> Original Paper: http://gist.cs.berkeley.edu/gisttalk.pdf
> High Performance Implementation: http://gist.cs.berkeley.edu/hiperf-gist.pdf

It won't work out of the box with MVCC concurrency for any number of
reasons.

> There is even a library with a compatible license.
>
>
>
>
>
>
>
>
> ------------------------------------
>
> Yahoo! Groups Links
>
>
>