#9598 closed defect (fixed)
Bogus "Areas share segment" warning by validator
Reported by: | mkoniecz | Owned by: | team |
---|---|---|---|
Priority: | normal | Milestone: | 14.02 |
Component: | Core validator | Version: | |
Keywords: | Cc: |
Description
Maybe there are cases where warning like this is useful but for now I encountered it only for buildings where it is correct tagging.
Example - entire old city of Kraków: http://www.openstreetmap.org/?mlat=50.061667&mlon=19.937222&zoom=15#map=16/50.0618/19.9371 (see https://en.wikipedia.org/wiki/File:Sukiennice_and_Main_Market_Square_Krakow_Poland.JPG and https://en.wikipedia.org/wiki/Main_Square,_Krak%C3%B3w)
Maybe after excluding building=* this rule will start reporting something useful.
Attachments (1)
Change History (15)
comment:1 by , 11 years ago
Milestone: | → 14.02 |
---|
comment:2 by , 11 years ago
comment:3 by , 11 years ago
Mmmm in that case we're talking about adjacent buildings that share a segment, yes, but converting all of them to multipolygons doesn't bring much. This is a normal way of tagging buildings in urban areas, I agree it should not raise any message, even informational.
Why do we have this check to begin with ?
comment:4 by , 11 years ago
Right, we could exclude building
s. In my area, some large forest areas are split at artificial points to make the way polygon smaller in size. Instead, they share a long+complex segment (e.g. way/146342814 + way/146342845
). This test identifies such cases.
comment:5 by , 11 years ago
We also could change this test to generate a warning only if a way segment consisting of at least 4 nodes is shared.
comment:6 by , 11 years ago
In France you can easily find buildings with small segments and more than 4 nodes. Think of buildings adjacent to churches for example, like here:
http://www.openstreetmap.org/#map=19/43.60096/1.43959
comment:9 by , 11 years ago
Replying to simon04:
Do we want a separate handing for buildings?
It is much better right now, but not still perfect in urban areas, there's still too much noise :) So yes I think we can ignore warnings when both areas are buildings.
comment:10 by , 11 years ago
Or maybe just when one of them is a building. There's plenty of areas drawn alongside a building (parkings, gardens, etc.). We could also ignore barriers (see #9695).
comment:12 by , 11 years ago
Resolution: | fixed |
---|---|
Status: | closed → reopened |
Is it possible to mention "it might be better to introduce multipolygons" in message? It is not obvious, I wanted to again report this bug because I had no idea what should be fixed.
I know that it would mean additional work for translators.
comment:13 by , 11 years ago
Resolution: | → fixed |
---|---|
Status: | reopened → closed |
Please don't reopen defects closed long ago to request something else. Please open a new enhancement for this, or simply update the online documentation for this warning.
If areas share a long segment, it might be better to introduce multipolygons. This is the idea of this informational message. Thus, I'd close this as
invalid
.