Skip to main content
Dear Mappers, 

 

 

During the LineOnAreaoverlayer, FME splits an object into two pieces with exactly the same attributes. Does anybody know why this would happen? It is the only object on the complete Belgian dataset behaving like this.

 

There exists indeed a polygon that uses the same edge, but that is the case for many of the line objects. Even the two line objects adjacent are topologically 'on top of' the area object.

 

 

Screenshots:

 

https://docs.google.com/file/d/0B3rvdWa8_6y-RmtxMUo2M0NEaEU/edit?usp=sharing https://docs.google.com/file/d/0B3rvdWa8_6y-YjBVSHlOdlI0NFk/edit?usp=sharing

 

 

 

best regards, 

 

Jelle

 

 

 

 
Hi,

 

 

it is difficult to see the polygons in the screenshots, it sounds like a possible rounding / precision issue. Try using an AnchoredSnapper (e.g. snapping the vertices to the polygon vertices) with a tiny tolerance before the LineOnAreaOverlayer.

 

 

You could also try a CoordinateRounder to see if that makes a difference.

 

 

David
Thanks for the help David.

 

 

There seems to be nothing wrong with the topology. The area and the line share the same edge and nodes, with the same coordinates. It should be, since the input is MultiNet.

 

I'll be pragmatic on this case and use the LineJoiner together with a ListIndexer on 0 for this particular object. It is probably the quickest workaround to this problem. I will have to check again when I process another dataset.

 

 

 

best regards, 

 

Jelle
in screenshot it looks like it might have beeen previously part or conneted to an arc? Maybe rounding issue therefore.

 

Or maybe it is the only line with no node on the intersection? Geometrically correct but topologically bad?

 

Aggregate sneeked in, noone noticed?

 


the useage notes for the lineonareaoverlayer say that

 

 

'if the lines and areas being overlaid have attributes with the same name, then the values will not be transferred from one to the other'

 

 

could there be a conflict with the attribute names? you could try the bulkattributerenamer to test this

Reply