[postgis-users] Question on topology

Neumann, Andreas a.neumann at carto.net
Wed Jul 13 04:05:41 PDT 2016


Hi Strk, 

Yes - I will try that next - loading it in chunks. But wouldn't it miss
out on some of the neighbourpolygons then if I use such subsets based on
pkey? 

Thanks,
Andreas 

On 2016-07-13 12:55, Sandro Santilli wrote:

> On Tue, Jul 12, 2016 at 01:15:41PM +0200, Neumann, Andreas wrote: 
> 
>> Hi Sandro, 
>> 
>> I let this function run during the night and after 7.5h I got an error
>> message:
> 
> Eh, that's why I suggested loading the topology in batches.
> 
> Use constructs like:
> 
> SELECT TopoGeo_addPolygon(...)
> FROM input_table
> WHERE gid >= $1 AND gid < $2
> 
> With $1..$2 in a range that makes it run within a few minutes.
> 
> Or, alternatively, code the loop in plpgsql by intercepting
> exceptions and skipping the "offending" input to analyze later
> (but I like the multi-transaction approach better as it allows
> you to see the topology in QGIS while it's being loaded).
> 
>> ERROR: Corrupted topology: adjacent edges 159972 and -159958 bind
>> different face (0 and 78697)
> 
> It means that somehow the ST_CreateTopoGeo function in a given
> stage created a corrupted topology. This is usually due to some
> robustness issue. Snapping may help a little with this.
> 
> Once again, note that if this was being done in a loop you might
> have stopped and looked at the issue with QGIS, to continue after
> the topology validity was fixed.
> 
>> I have to say that I did not tune Postgis memory-wise - maybe I should
>> do that next. But the above error message probably doesn't indicate a
>> memory problem, but a problem with the data.
> 
> Correct, it's not a problem with memory.
> 
>> My data: 2582 polygons with a total of 167176 vertices.
> 
> Try loading it in chunks of 300 polygons
> 
> --strk;
> 
> ()   Free GIS & Flash consultant/developer
> /\   https://strk.kbt.io/services.html

  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20160713/fc23b1dd/attachment.html>


More information about the postgis-users mailing list