House number range shown in OSM but not in Osmand

[Stephan Marx]

Hi,

based on the 2 maps North Rhine-Westphalia and Germany adresses-nationwide I tried to find a specific House no. in the Osmand Adress search.
The House is:House no 25, Street Lüttelforst, Town: Lüttelforst (you need to use the small city search).

While the map itself  in Osmand and OSM shows the House (and all others in the street), the House # selector in Osmand only offers House no. 94 to 140.

Anyone with a clue what went wrong here?

BR
Stephan

In OSM this is (applicable for all # between 1 and 93):

Node: 1734958329

Import der Adressen Kreis Viersen 5

Edited about 3 years ago by Shape_Import_Viersen

Version #1 · Changeset #11447065

Location: 51.1810708, 6.2748002

Tags

addr:city	Schwalmtal
addr:country	DE
addr:housenumber	25
addr:postcode	41366
addr:street	Lüttelforst
source	Kreis_Viersen_Katasteramt_2012_03

[Stephan75]

Hello Stephan ;-)

this is a maybe well known problematic of Osmand's address database.

The best OSM data is a political boundary for a place, in this case http://www.openstreetmap.org/relation/88529

Osmand uses this boundary for address index if there is an object inside with place=* AND name=Schwalmtal

In this case there is no such single node with place=* AND name=Schwalmtal

Thus you even cannot find "Schwalmtal" in Osmand's list of places.

For the hamlet Lüttelforst it seesm that we have NO boundary relation where Osmand can determine all streets belonging to Lüttelforst.

So Osmand mapcreator seems to use a kind of estimation to find all housenumbers belonging to Lüttelforst.

Some of them are in, some are missing.

Stephan

[Kurt Waldhans]

Hallo Stephan75,
do you know exactly what data are missing in OSM?. We are trying keeping the addresses in Kreis Viersen up-to-date.

I have seem the problem, so maybe we can solve it.

Since Schwalmtal is working, I checked the differences.

 In the Schwalmtal Relation 88686  "place=town" is set.

Shoul I tag this to the other admin relations? This could solve the same problem in 'Tönisvorst'

[Stephan Marx]

Thanks for the answer Stephan75, fixing this in OSM might for me to be a bit more learing effort in terms of Boundaries and relations (so far I limited myself to smaller elements).
But maybe Kurt in the following Post has some better expertise to fix this issue.

BR
Stephan

[Jack Burke]

I didn't realize that Osmand needed s place= tag for some of those in-between-town nodes.

Fron my own testing, it appears to also recognize the is_in= tag structure, but does anyone know for certain that it does?

-jack

[Harry van der Wolf]

As far as I know the is_in is not used by OsmAnd. It is a (strongly) deprecated tag anyway.

All address handlings is done here: https://github.com/osmandapp/OsmAnd-tools/tree/master/OsmAndMapCreator/src/net/osmand/data/preparation/address

The strange thing is though that the POI edit functionality of OsmAnd does support it: https://github.com/osmandapp/Osmand/tree/master/OsmAnd-java/src/net/osmand/osm/edit

[Kurt Waldhans]

I have added 'place=town' to all 'admin_level=8' relation inside Kreis Viersen. Let's way for the next map updated and check the adresse search again.

[Stephan Marx]

Thanks Kurt .. will have a look after the next Osmand Update!

Gruß

Stephan

2015-06-15 0:08 GMT+02:00 Kurt Waldhans <kurtwa...@gmail.com>:

I have added 'place=town' to all 'admin_level=8' relation inside Kreis Viersen. Let's way for the next map updated and check the adresse search again.

--
You received this message because you are subscribed to a topic in the Google Groups "Osmand" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/osmand/RiazDnPUH9w/unsubscribe.
To unsubscribe from this group and all its topics, send an email to osmand+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Stephan75]

Hello Kurt,

Years ago I also added some place=XXX nodes for boundaries where no equal-named node was existant, althoug there is no real single place with that in real world ... examples are Rosengarten and Seevetal in the south of Hamburg in Germany.

There is also an overpass-turbo query to find boundary relations where no place-node with same name is present in OSM database, see my OSM wiki userpage

http://wiki.openstreetmap.org/wiki/User:Stephan75

(edit that query with any Bundesland in Germany or any other province around the world ... in case of error try to increase timeout.

BUT advice to all:

There are some mappers who consider setting place nodes for places that do not exist in reality only because Osmand needs them is "Tagging for the Renderer" ... in this case Osmand and its address search.

Stephan

[Kurt Waldhans]

The problem with address search seem to be more widespread, see topic "somes villages are not available in the search by address" and "OsmAnd cannot find city".

Can we document what Osmand needs for a reasonable address search? Or does such a document exist?

PS: "Tag for Renderer": let's see, I wrote explicit "address search inside Osmand" into the changeset. If it works, I don't see a problem defending this tag.

[Harry van der Wolf]

2015-06-15 19:25 GMT+02:00 Stephan75 <der.steph...@googlemail.com>:

BUT advice to all:

There are some mappers who consider setting place nodes for places that do not exist in reality only because Osmand needs them is "Tagging for the Renderer" ... in this case Osmand and its address search.

Stephan

Hi Stephan,

Yes, this is indeed mentioned, but most of the time they mean changing existing correct (!) tagging, because some application can't deal with it. That's something else then simply adding missing data into OSM even though (perhaps) only OsmAnd is using it.

Harry

 

[Stephan75]

Hello Kurt,

as long as we can see it from googlecode server, see

https://code.google.com/p/osmand/wiki/MapAddressDataStructure ... but maybe a bit out of date already.

PS: What about transferring all that stuff to github wiki pages of Osmand?????

Stephan