Modify

Opened 14 years ago

Closed 10 years ago

#4582 closed defect (fixed)

Overlapping Areas misreported

Reported by: pinkduck Owned by: team
Priority: major Milestone:
Component: Core validator Version: tested
Keywords: overlapping area way Cc: pinkduck, AM909, landwirt@…, moltonel, scai

Description (last modified by simon04)

With two closed ways representing two mutually exclusive areas, if both are separate there is no issue. If they share a single node at a vertex there is no validation report. However, if they share two nodes of a common edge then an 'Overlapping areas' validation 'Other'-type notification is given by the JOSM Validator plug-in.

When these two areas only overlap at their polygon borders I don't think there should be a validation notification raised. While there is a one-dimensional overlapping line of the virtual separation of a 2D surface, it tends towards zero given an infinitesimal approximation.

If there is any true overlap of the areas defined by the polygonal borders then this 'Other' notification is acceptable. This could be calculated using an algorithm similar to that described at: http://mathforum.org/library/drmath/view/63171.html. However, with the adaptation that parallel test edges should be excluded from determining the overlap. An obvious exception being when all nodes of each area overlap.

Attachments (3)

4582_no_overlap.osm (1010 bytes ) - added by simon04 13 years ago.
4582_still_problem.osm (16.2 KB ) - added by skyper 12 years ago.
example osm file
true_overlapps.osm (13.2 KB ) - added by skyper 12 years ago.
second try, forgot to save before upload

Download all attachments as: .zip

Change History (31)

comment:1 by pinkduck, 14 years ago

Cc: pinkduck added
Version: latest

comment:2 by AM909, 14 years ago

Cc: AM909 added
Priority: normalmajor
Version: latesttested

Agreed. In the US, administrative boundaries are quite commonly, correctly, glued together. The warning should only occur if they actually cross - not if they simply share edges.

Another example is landuse polygons. It is common, and correct, for housing tracts to be glued together along their interfaces, and this should not result in an error.

Too many false positives, and warnings are ignored.

comment:3 by e2jk, 14 years ago

Ticket #4768 has been marked as a duplicate of this ticket.

comment:4 by e2jk, 14 years ago

From Ticket #4768:

@skyper:
If two areas with landuse=* area bound together using the same nodes as boundery validator should not inform about overlapping areas. The same for buildings and ways as areas. For example a parking-area with highway=service and a building.

But it should inform or even warn if these area really overlap which it does not !!

@e2jk:
This also happens for waterways that are connected (but not really overlapping)

comment:5 by e2jk, 14 years ago

Type: enhancementdefect

comment:6 by skyper, 13 years ago

This check is not much useful right now, as it does not work for really overlapping areas and only leads to use relations for all areas which have at least one border sharinging with another area which is the default !

comment:7 by simon04, 13 years ago

In [4448/josm]:

see #4582 - validator test for overlapping areas

by simon04, 13 years ago

Attachment: 4582_no_overlap.osm added

comment:8 by simon04, 13 years ago

Validator test added – please test and report :-).

Geometry.polygonIntersection has a problem with a certain type of overlap (cf. attachment:4582_no_overlap.osm).

by skyper, 12 years ago

Attachment: 4582_still_problem.osm added

example osm file

comment:9 by skyper, 12 years ago

Still have misreported overlaps which are only sharing borders. Have a look at 4582_still_problem.osm.

Problems are with buildings and boundary.

r4475

Last edited 12 years ago by skyper (previous) (diff)

comment:10 by simon04, 12 years ago

Right, I only removed 1 out of 4 such errors. Currently, one will get "Overlapping {highways,railways,ways} (with area)".

Apart from the wrong wording (should be something like "{highway,railway,way} is sharing segment with area"), the main question is whether this informational warning is worthless at all. I tend to drop this.

Please comment :-).

in reply to:  10 comment:11 by skyper, 12 years ago

Replying to simon04:

Right, I only removed 1 out of 4 such errors. Currently, one will get "Overlapping {highways,railways,ways} (with area)".

Major improvement would be if Josm finds overlapping areas which it still does not ! See comment:8

Apart from the wrong wording (should be something like "{highway,railway,way} is sharing segment with area"), the main question is whether this informational warning is worthless at all. I tend to drop this.

Please comment :-).

Please, don't drop it. Most of the times, I consider these warnings as errors or at least areas to investigate.

comment:12 by simon04, 12 years ago

Major improvement would be if Josm finds overlapping areas which it still does not ! See comment:8

In your example, which overlap should be reported, i.e., between which objects?

Please, don't drop it. Most of the times, I consider these warnings as errors or at least areas to investigate.

Thus, this informational warning should remain but with a different wording (see suggestion above)?

Last edited 12 years ago by simon04 (previous) (diff)

in reply to:  12 comment:13 by skyper, 12 years ago

Replying to simon04:

Major improvement would be if Josm finds overlapping areas which it still does not ! See comment:8

Your example shows that a overlap between a way and an area isn't reported.

Sorry, I think I got it now:

The way is a boundary and part of a multip. relation. Maybe leave out ways in general and only look for some (highway,waterway...)

Please, don't drop it. Most of the times, I consider these warnings as errors or at least areas to investigate.

Thus, this informational warning should remain but with a different wording (see suggestion above)?

+1

comment:14 by simon04, 12 years ago

For consistency, the remaining test errors of this kind, i.e., "Overlapping {highways,railways,ways}", should also renamed to "{highways,railways,ways} share segment"?

Last edited 12 years ago by simon04 (previous) (diff)

in reply to:  12 comment:15 by skyper, 12 years ago

Replying to simon04:

Did miss that:

In your example, which overlap should be reported, i.e., between which objects?

There exists no overlap, only if you count the multip.-relation as area. All other are only touching!

My example is not very good, maybe I can add a better one.

  • The test should not include ways with tag boundary.
  • does the check work with multipolynom relations ?

comment:16 by simon04, 12 years ago

In [4496/josm]:

see #4582 - clarify warning messages of overlapping way validator test

in reply to:  14 ; comment:17 by simon04, 12 years ago

Description: modified (diff)

Replying to simon04:

For consistency, the remaining test errors of this kind, i.e., "Overlapping {highways,railways,ways}", should also renamed to "{highways,railways,ways} share segment"?

This question is to be decided before closing this ticket.

in reply to:  17 comment:18 by skyper, 12 years ago

Replying to simon04:

Replying to simon04:

For consistency, the remaining test errors of this kind, i.e., "Overlapping {highways,railways,ways}", should also renamed to "{highways,railways,ways} share segment"?

This question is to be decided before closing this ticket.

It does not even find all overlapping buildings. I have creating a test files, which needs to be improved but it is a start.

by skyper, 12 years ago

Attachment: true_overlapps.osm added

second try, forgot to save before upload

comment:19 by skyper, 12 years ago

Ticket #7236 has been marked as a duplicate of this ticket.

comment:20 by skyper, 12 years ago

Cc: landwirt@… added

comment:21 by skyper, 10 years ago

Ticket #9140 has been marked as a duplicate of this ticket.

comment:22 by skyper, 10 years ago

Cc: moltonel added
Keywords: area way added; areas removed

comment:23 by scai, 10 years ago

Cc: scai added

comment:24 by Don-vip, 10 years ago

In 6300/josm:

see #4582, see #9140 - Fix validator false positive "overlapping ways" for adjacent waterway=riverbank

comment:25 by Don-vip, 10 years ago

In 6301/josm:

see #4582, see #7236 - Fix validator false positive "overlapping ways" for adjacent multipolygons

comment:26 by Don-vip, 10 years ago

In 6304/josm:

see #4582, fix #8362 - Fix validator false positive "overlapping ways" for building:part

comment:27 by moltonel, 10 years ago

The fix works for me (svn 6312), at least for the waterway usecase. Thanks.

comment:28 by Don-vip, 10 years ago

Resolution: fixed
Status: newclosed

Closing this 4-year old ticket as I believe I have fixed the reported examples. Please open a new one with up-to-date description and samples if you find other false positives.

Modify Ticket

Change Properties
Set your email in Preferences
Action
as closed The owner will remain team.
as The resolution will be set.
The resolution will be deleted. Next status will be 'reopened'.

Add Comment


E-mail address and name can be saved in the Preferences .
 
Note: See TracTickets for help on using tickets.