Re: [OSM-dev] Polygon inner/outer relation in osm file

[Paul Norman]

On 8/23/2016 6:58 AM, Martin Koppenhoefer wrote:

recently we stumbled upon problematic behavior of current multipoligon 
processing in osm-carto


OpenStreetMap Carto does not do any multipolygon processing. That is all 
done by osm2pgsql, and the behavior of osm2pgsql has changed 
significantly from the version used on the OSMF tile servers.


___
dev mailing list
dev@openstreetmap.org
https://lists.openstreetmap.org/listinfo/dev

Re: [OSM-dev] Polygon inner/outer relation in osm file

[Jochen Topf]

On Di, Aug 23, 2016 at 03:58:02 +0200, Martin Koppenhoefer wrote:
> > Il giorno 23 ago 2016, alle ore 15:21, Frederik Ramm  
> > ha scritto:
> > 
> > they can have missing or wrong inner/outer tags
> > (usually considered valid)... the C++/Python library "osmium" is
> > probably the most advanced in building proper polygons, or check out
> > osm2pgsql, imposm, or ogr2ogr which also have code to deal with that.
> 
> 
> recently we stumbled upon problematic behavior of current multipoligon 
> processing in osm-carto, as it drops tags on inner relation members if the 
> outer members (or the relation?) has the same tags. I believe this kind of 
> "fix" should not be performed, it leads to problems if this kind of mapping 
> was done on purpose, e.g. because of some properties being different.
> 
> Not applying "magic guesswork" also leads to more predictable behavior and 
> ultimately to mappers fixing the representation in case it was wrong.

This has been a long-standing problem. Unfortunately there is a lot of
data in OSM that is still tagged this way, so it is not an option to
just not do this anymore. Too many multipolygons (about 17,000) would
break. This is not the only problem with current multipolygon tagging,
and I am working on a "campaign" to get these fixed.

There is more background and some stats here:
https://github.com/osmlab/fixing-polygons-in-osm
http://area.jochentopf.com/stats/

Jochen
-- 
Jochen Topf  joc...@remote.org  http://www.jochentopf.com/  +49-351-31778688

___
dev mailing list
dev@openstreetmap.org
https://lists.openstreetmap.org/listinfo/dev

Re: [OSM-dev] Polygon inner/outer relation in osm file

[Martin Koppenhoefer]

sent from a phone

> Il giorno 23 ago 2016, alle ore 15:21, Frederik Ramm  ha 
> scritto:
> 
> they can have missing or wrong inner/outer tags
> (usually considered valid)... the C++/Python library "osmium" is
> probably the most advanced in building proper polygons, or check out
> osm2pgsql, imposm, or ogr2ogr which also have code to deal with that.


recently we stumbled upon problematic behavior of current multipoligon 
processing in osm-carto, as it drops tags on inner relation members if the 
outer members (or the relation?) has the same tags. I believe this kind of 
"fix" should not be performed, it leads to problems if this kind of mapping was 
done on purpose, e.g. because of some properties being different.

Not applying "magic guesswork" also leads to more predictable behavior and 
ultimately to mappers fixing the representation in case it was wrong.


cheers,
Martin 
___
dev mailing list
dev@openstreetmap.org
https://lists.openstreetmap.org/listinfo/dev

Re: [OSM-dev] Polygon inner/outer relation in osm file

[Frederik Ramm]

Hi,

On 08/23/2016 09:39 AM, patrick keshishian wrote:
> What is the mechanism used, while processing "relation"-s, in
> matching "inner" and "outer"-s of polygons?

There's no guarantee that role="inner" and role="outer" are even set. I
strongly advise against trying to write your own algorithm that combines
polygons; you will get 80% right quickly and fight with the other 20%
for weeks. Rings can be nested (valid), self-intersecting (invalid) or
missing bits (invalid), they can have missing or wrong inner/outer tags
(usually considered valid)... the C++/Python library "osmium" is
probably the most advanced in building proper polygons, or check out
osm2pgsql, imposm, or ogr2ogr which also have code to deal with that.

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frede...@remote.org  ##  N49°00'09" E008°23'33"

___
dev mailing list
dev@openstreetmap.org
https://lists.openstreetmap.org/listinfo/dev

Re: [OSM-dev] Polygon inner/outer relation in osm file

[Michael Zangl]

Hi,

There is no order for any of the multipolygon members. When I create them,
I try to sort them (outer before inner, then to form circles). JOSM
combines them to form circles for some months now. But there is no
convention.

So it is up to you to match the corresponding outer/inner rings and
find out if the multipolygon is valid.

> http://wiki.openstreetmap.org/wiki/Relation:multipolygon#Valid_Multipolygon_conditions

Michael

Am 23.08.2016 um 09:39 schrieb patrick keshishian:
> Greetings,
> 
> What is the mechanism used, while processing "relation"-s, in
> matching "inner" and "outer"-s of polygons?
> 
> Examining some example data extracts (from geofabrik.de) it looked
> as if an "outer" is followed by its "inner"-s (if any).
> 
> But soon I came across ones like this:
> 
>changeset="17984334" uid="574654" user="Tom_Holland">
>   
>   
>   
>   
>   
>   
>   
>   
>   
> 
> 
> I am really hoping to be missing something obvious, and that the
> import (or render) software does not examine each "inner" against
> each "outer" to determine the association.
> 
> Also hoping this is the appropriate list for this sort of questions.
> 
> --patrick
> 
> [0] http://www.openstreetmap.org/relation/3215648#map=16/19.9202/-155.8783
> 
> ___
> dev mailing list
> dev@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/dev
> 


___
dev mailing list
dev@openstreetmap.org
https://lists.openstreetmap.org/listinfo/dev

[OSM-dev] Polygon inner/outer relation in osm file

[patrick keshishian]

Greetings,

What is the mechanism used, while processing "relation"-s, in
matching "inner" and "outer"-s of polygons?

Examining some example data extracts (from geofabrik.de) it looked
as if an "outer" is followed by its "inner"-s (if any).

But soon I came across ones like this:













I am really hoping to be missing something obvious, and that the
import (or render) software does not examine each "inner" against
each "outer" to determine the association.

Also hoping this is the appropriate list for this sort of questions.

--patrick

[0] http://www.openstreetmap.org/relation/3215648#map=16/19.9202/-155.8783

___
dev mailing list
dev@openstreetmap.org
https://lists.openstreetmap.org/listinfo/dev