[postgis-devel] New node splitting algorithm for GiST

Paul Ramsey pramsey at opengeo.org
Mon Feb 20 07:36:48 PST 2012


On Fri, Feb 17, 2012 at 12:41 AM, Mark Cave-Ayland
<mark.cave-ayland at siriusit.co.uk> wrote:
> On 16/02/12 17:37, Paul Ramsey wrote:
>
>>> In terms of getting a 2.0 release out, I don't think this is the right
>>> time
>>> to be doing this for such a new feature. I'd much rather have a shorter
>>> 2.1
>>> cycle (say 6 months) and apply the patch just after branch to give it the
>>> testing time it requires rather than add more complicated #ifdeffery in a
>>> critical piece of code.
>>
>>
>> I really don't see the harm in having an EXPERIMENTAL_INDEX block,
>> particularly since it will be (should be, can be) one big block, not
>> scattered throughout the codebase. Gives us a chance to get some
>> testing without waiting through another major release cycle, which as
>> we've seen... can be long? :)
>
>
> Well, what testing? If it's an experimental feature, it's going to be
> disabled by default, right? So therefore it's not going to be included by
> any of the package distributors meaning the only people that will test it
> will be developers... and they will be using SVN anyhow so I don't see your
> point?
>
> I'd much rather take a bit longer to work with Alexander and come up with a
> better algorithm that can replace the old one entirely and test it early in
> the 2.1 cycle, rather than concocting another convoluted set if #ifdefs
> which are going to cause yet more maintenance and regression pains when we
> are trying to get 2.0 out of the door...

I think you're overstating the complication of this, it's just a
replacement for picksplit. It's one API piece, so the whole thing can
be one #define. And as far as improvement goes, Alexander has already
shown he can produce good results in the lab, the only thing missing
seems to be real-world verification that his method is better across a
range of data, which is something that having his implementation
optionally available can provide to folks who want to live on the
edge.

I remain +1.

P.

>
>
> ATB,
>
> Mark.
>
> --
> Mark Cave-Ayland - Senior Technical Architect
> PostgreSQL - PostGIS
> Sirius Corporation plc - control through freedom
> http://www.siriusit.co.uk
> t: +44 870 608 0063
>
> Sirius Labs: http://www.siriusit.co.uk/labs
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-devel



More information about the postgis-devel mailing list