[postgis-users] More details for errors

[Denis Rouzaud]

Hi again!

Well, this is always the same problem, if I do linemerge(union()), I 
have to force multi to prevent errors. Then, simplifying to linestring 
seems to consume resources.

On way, would to do a view which determine if the union would lead to a 
multi or a linestring.
But I don't know how to achieve this.
I have no idea either if this would also be greedy on resources.

Denis

On 03/27/2012 05:43 PM, Nicolas Ribot wrote:
>> Yes, it is 2.0
>> http://postgis.refractions.net/documentation/manual-svn/ST_CollectionHomogenize.html
>>
>> Well, I am turning around the problem...
>> I have to do a cast, as I used
>> ST_Multi(ST_LineMerge(ST_Union(wkb_geometry)))::geometry(MultiLineString,xxx)
>> in the first view to prevent errors of unconnected lines.
>>
>> Right now the best solution I have is to let the view being unusable, and
>> another view to report errors.
>> This is not very satisfying as if a problem occurs, I cannot use my view
>> until I fix the geometric errors.
>>
>> Basically, I need to compute a view with a LineMerge returning LineString
>> (i.e. leaving away features leading to MultiLineString) and another view
>> returning IDs of features leading to MULTI.
>>
>> I thought about using ST_Touches but it only uses two geometries and I can
>> have more in the LineMerge. No idea what to look for...
>>
>> Thanks for your help. I could go deeper in the problem!
>>
>> Greetings
>>
> I would then create 3 views:
> One doing the actual computation (linemerge(Union)), then one view
> where st_numgeometries(the_geom) = 1, and one view where
> st_numgeometries(the_geom)>  1.
>
> nicolas
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users