Hello!
Some polygons disappear in the transformer dissolver, it doesn´t matter if I use group by or not.
Coordinaterounder, Densifier, Snapper fix some problem but other polygons disappear instead.
Geometry validator with no result.
Tips or idea?
Userlevel 6
+32
- Celebrity
- 2339 replies
-
22 December 2016
What exact version of FME?
Can you share the geometry so we can reproduce the error? Just strip the attributes and write to FFS.
You can easily anonymize the features with an offsetter and rotator if needed.
+16
- Contributor
- 1425 replies
-
22 December 2016
The all point of the Dissolver is to dissolve polygons and therefore make some 'disappear', can you visualize your unexpected results and share it?
Here is a screenshot and input file for this example...
small-example3-mergeid.zip
Try to dissolve with group by for the attribute mergeid...
+16
- Contributor
- 1425 replies
-
22 December 2016
Here is a screenshot and input file for this example...
small-example3-mergeid.zip
Try to dissolve with group by for the attribute mergeid...
My guess is that there are inconsistencies between the features, especially mis-alignement of nodes, possibly using a coordinaterounder before the dissolver might help.
What exact version of FME?
Can you share the geometry so we can reproduce the error? Just strip the attributes and write to FFS.
You can easily anonymize the features with an offsetter and rotator if needed.
FME 2016.1 - 32bitFME 2015.0 - 32bit
FME 2015.1.1 - 64bit
+16
- Contributor
- 1425 replies
-
22 December 2016
It works sometimes with coordinaterounder. But polygons disappear elsewhere. Very frustrating!
Userlevel 4
- 8178 replies
-
3 January 2017
Is it possible that some of your polygons contains arcs?
I've seen quite a lot of similar bugs on datasets containing arcs, probably due to floating point precision issues.
Consider pre-processing your data with an ArcStroker and a Snapper (segment snapping) before the Dissolver.
Userlevel 4
+13
- Contributor
- 3726 replies
-
4 January 2017
@sjodin
I can advise that we made massive improvements in this area in FME 2017. Please try the original dissolves with FME 2017 beta and let us know at support@safe.com if there are still issues. @david_r - you'll be pleased to know that the improvements also included robustness in arc situations, so there should be no need for stroking and snapping prior to dissolve.
Userlevel 4
- 8178 replies
-
4 January 2017
@sjodin
I can advise that we made massive improvements in this area in FME 2017. Please try the original dissolves with FME 2017 beta and let us know at support@safe.com if there are still issues. @david_r - you'll be pleased to know that the improvements also included robustness in arc situations, so there should be no need for stroking and snapping prior to dissolve.
+7
@sjodin
I can advise that we made massive improvements in this area in FME 2017. Please try the original dissolves with FME 2017 beta and let us know at support@safe.com if there are still issues. @david_r - you'll be pleased to know that the improvements also included robustness in arc situations, so there should be no need for stroking and snapping prior to dissolve.
We have also experienced quite some issues with polygons disappearing when using the dissolver. Looking forward for FME 2017!!
+3
Hi,
Both areas in the sample have huge spikes...
If you use spikeremover at max 1 degree, the output will dissolve correctly.
Userlevel 4
+13
- Contributor
- 3726 replies
-
4 January 2017
That's great news!
We have also experienced quite some issues with polygons disappearing when using the dissolver. Looking forward for FME 2017!!
Reply
Enter your username or e-mail address. We'll send you an e-mail with instructions to reset your password.