[postgis-devel] Bugs in gserialized_gist_compress_2d

[Alexander Korotkov]

On Thu, Sep 26, 2013 at 11:35 PM, Alexander Korotkov
<aekorotkov at gmail.com>wrote:

> On Thu, Sep 26, 2013 at 3:00 AM, Paul Ramsey <pramsey at cleverelephant.ca>wrote:
>
>> OK, I've ticketed it, for what that's worth. I do believe it's an
>> important correctness issue.
>>
>> http://trac.osgeo.org/postgis/ticket/2487
>>
>> Since I have you here... any guesses on this?
>>
>> http://obartunov.livejournal.com/174411.html
>
>
> This benchmark is about geography indexing of which is defined in
> gserialized_gist_nd.c. I've touched only gserialized_gist_2d.c before.
> I supposed geography GiST would be much slower than pgSphere, but 2 orders
> of magnitude surprises me. It should be not only bad splitting algorithm,
> but also some bugs among opclass.
>

Another point you should note is that tests were done on quite big amount
of points: 10^8. Such test can discover defects that aren't visible on
lower amount of data.

------
With best regards,
Alexander Korotkov.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20130927/43bf6c8e/attachment.html>