[postgis-devel] a slight error in postgis TopologySummary

Rémi Cura remi.cura at gmail.com
Thu Sep 26 08:00:28 PDT 2013


hey,
thanks for the answer.
I'm not confused (or non more than usual :-) ).
Postgres sequences are another matter, I personally find non-successing
number not optimal because it makes thinks less reproducible. (I'm in a
state S. I create A then I destroy A. Yet I'm not in state S ). I agree it
greatly simplifies things and I use it too.


It is a very small bug, and I'm absolutely not criticizing or lecturing :
I'm a beginner.

It was a flaw in design : to get all the layer on summary we shouldn't look
into relations table (too low level : edge & co), because a layer can exist
and have 0 relation.
(and for performance it is also a bad idea to look into big table and do
distinct ...)

On the opposite we should look into a high level table (like
topology.layer) to get the information.

Anyway it's a very small problem that could be stated like this :
when using summary on a topology where there is an empty layer (meaning it
has yet no topogeom inside), this layer is not counted in summary :
As you can see on the output, I have 3 layers, yet summary count 2.


Topology demo_zone_test (11), SRID 931008, precision 0.1, has Z
88 nodes, 105 edges, 49 faces, 76 topogeoms in  2 layers
Layer 1, type Lineal (2), 48 topogeoms
 Deploy: demo_zone_test.route_demo.tg
Layer 4, type Polygonal (3), 28 topogeoms
 Deploy: demo_zone_test.bati_demo.tg
Layer 5, type Polygonal (3), 0 topogeoms
 Deploy: demo_zone_test.chaussee.tg

Anyway what you propose is equivalent ( If there is no more than 1 topology
per schema, but I think this is the case by construction), so do as you
please. I was trying to imitate your way of coding to blend in.

I'm stopping to work on topology for the moment anyway as it is very
unstable in this version, so I'll install the new version before going on.

Cheers,

Rémi-C
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20130926/2600ce8f/attachment.html>


More information about the postgis-devel mailing list