[postgis-users] Topology Postgis 2.0

[Wim Nederend]

On 09/27/2012 02:50 PM, Sandro Santilli wrote:
> On Thu, Sep 27, 2012 at 02:44:12PM +0200, Sandro Santilli wrote:
>> On Thu, Sep 27, 2012 at 04:58:24AM -0700, ps.tiara wrote:
>
>>> but when i visualized the result of topology in x3d, it was mess..maybe i
>>> made a mistake by wrong query??
>>
>> How did you visualize ? What do you mean by "messy" ?
>> The topology in your backup results valid, and I can see it just fine.
>> But yes, I do see a lot of very small faces, a lot of edges close
>> to one other.
>
> One more tip, you can "see" how many "messy" faces you have with this
> query:
>
> with inp as (
>   SELECT face_id, st_area(ST_GetFaceGeometry('simpanglima_topo', face_id))
>   FROM simpanglima_topo.face where face_id > 0
> ) SELECT * from inp order by st_area;
>
> You'll see there are only 4 faces with an area > 10 square units, while
> 92 faces have an area < 0.0000006 square units...
>
> It's still topologically valid, but probably not what you want.
> It'd be an interesting research topic (since you're doing a research)
> to figure out HOW to fix that situation _now_ that you have it encoded
> as a topology.
>
> I think you'll basically want to delete each face and make it part
> of one of the big ones. You could recursively find an edge shared
> between a big face and a small face and remove it.
>
> --strk;

The workaround I described in ticket 
http://trac.osgeo.org/postgis/ticket/1946 helped for me.

Regards, Wim