Can Validator pick up paths connected to land use boundaries ?

[anonymous]

Several times recently I have incorrectly connected a minor path (footway, cycleway etc.) to a land use boundary rather than the intended adjacent path or road. I have only noticed either this when rendering shows the error or on later review for another reason. Is the end of a path at a land use boundary something Validator could/should detect and report on ?

[Harry Wood]

Yes I think this is a common problem with roads as well as footways.

I know a lot of mappers deliberately connect footways to landuse areas as well as continuing it to the adjacent path/road. So the connection itself is not an error, but maybe a if the footpath terminates at the connection with landuse, this could count as a warning. Likewise for landuse-like tags such as leisure=park.

[delta_foxtrot2]

use noexit=yes tag to indicate a dead end and that node will be ignored with recent patches I've submitted.

[simon04]

Patch attached:

• adds utility functions
  • AbstractPrimitive: boolean hasTag(String key, String... values)
  • ElemStyles: AreaElemStyle getAreaElemStyle(OsmPrimitive p, boolean pretendWayIsClosed), boolean hasAreaElemStyle(OsmPrimitive p, boolean pretendWayIsClosed)
  • Way: Node getFirstNode(), Node getLastNode()
• adds validation test WayConnectedToArea

Please take a look at the provided patch!

[stoecker]

Way has already functions firstNode() and lastNode(). No need to reimplement them.

[simon04]

In [4682/josm]:

fix #2746 - add validation: Way connected to Area

[simon04]

Replying to stoecker:

Way has already functions firstNode() and lastNode(). No need to reimplement them.

Right… Changed and committed.

[Don-vip]

There are plenty of false positives with this new warning, like a highway ending on a roundabout, a footway ending on a building entrance... This check should be be only restricted to landuse boundaries right now, and extended to other specific cases if needed.

[simon04]

In [4704/josm]:

fix #2746 - improve validation: Way connected to Area

[Don-vip]

Sorry to reopen again this ticket :)
I have another case where this warning should not happen: when a landuse share a way segment with a highway A, and another highway B connects to A:

            _____ highway A
           /
           |
-----------|
|          |
| landuse  |-------- highway B
|          |
-----------|
           |

This is perfectly valid :)

[abienvenu]

Replying to Don-vip:

Sorry to reopen again this ticket :)
I have another case where this warning should not happen: when a landuse share a way segment with a highway A, and another highway B connects to A:

            _____ highway A
           /
           |
-----------|
|          |
| landuse  |-------- highway B
|          |
-----------|
           |

This is perfectly valid :)

Yes Don-vip, it is valid, and apparently is has been fixed somewhere before version 5836!

However, there is still an issue: if some highway shares a way with landuse, and this highway is split in two segments (because it has different properties, like maxspeed), we get two "Way connected to Area" warnings at the connection point.

The validation should not check if a way is connected to an area, but should check if a way end is connected to nothing else than an area (and noexit is not set to "yes" on that end node).

[Don-vip]

Replying to abienvenu:

However, there is still an issue: if some highway shares a way with landuse, and this highway is split in two segments (because it has different properties, like maxspeed), we get two "Way connected to Area" warnings at the connection point.

Seems to have been fixed now, cannot reproduce with latest. If you find another false positive, feel free to open a new ticket :) Closing this one for good !

[Don-vip]

Ticket #4682 has been marked as a duplicate of this ticket.

[GerdP]

see #19600 for further thoughts/problems