[postgis-users] speeding up simple pt-in-poly lookups

[Mr. Puneet Kishor]

On Dec 22, 2011, at 11:09 PM, Martin Davis wrote:

> That's what I meant by "size'.  That's quite a difference!  And would definitely explain the performance improvement.
> 
> One other thing I'm curious about.  Is the coordinate system is lat/long?  In which case the buffer distance of 0.75 is quite large, isn't it?  Is that going to possibly produce incorrect results?


Yes, it is lat/lng, and yes, the buffer is rather large. But I viewed the buffered geometry in QGIS, and it is fine for my work. It coalesces a few islands here and there making my map look like the Michelin tire man, but I don't care about the outline. I just want to grab the points, and, for now, speed is winning over too-much-accuracy. The nice thing with this approach is, I don't have to create a new geometry (which I would have to do if I were to chop up the continent), so I can experiment with different simplify and buffer tolerances till things are correct.


> 
> On 12/22/2011 9:04 PM, Mr. Puneet Kishor wrote:
>> 
>> What do you mean by "size"? If you mean "number of points," the original geometry (North America continent) had circa 50K points, and the simplified geometry had around 2000, so an order of magnitude difference. I am speaking from memory because I am not in front of my db server. Will check again tomorrow, and repost if the numbers are drastically different from above.
>> 
>> 
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users