Hi Gerd
Yes, there are many inconsistencies there, back issues before interpolation.
What did you do at the point there, it was correct because it was exactly in
line.
I also see that there is inconsistency.
The POI, this is typically placed at the actual point and is searchable on GPS.
What is not generally POI when you do an interpolation, using meters, this
will come close to the real, but sometimes there may be inconsistencies.
What is not POI, thinking about this GPS has an address with number.
Soon this will incur each type of culture as is done this house numbering.
But today, getting the average of 40 meters that left her, already solve much
of the problem, how it is evolving, tomorrow may arise something to improve it.
_______________________________________________________________________________________________________
Anor C. A. de Souza
Concórdia SC
From: [email protected]
To: [email protected]
Date: Sat, 29 Aug 2015 08:55:59 +0200
Subject: Re: [mkgmap-dev] Commit: r3635: corrections to addr:interpolation way
checks
Hi Anor,
yes, that's right, but I see other errors in this area, for example
WARN: ...osmstyle.housenumber.HousenumberGenerator f:\osm\anor-carlos.osm:
addr:interpolation way http://www.openstreetmap.org/way/349694725 703..1305,
step=2 is ignored, it produces duplicate number(s) too far from existing nodes
WARN: ...osmstyle.housenumber.HousenumberGenerator f:\osm\anor-carlos.osm:
addr:interpolation way http://www.openstreetmap.org/way/349694729 1455..2151,
step=2 is ignored, it produces duplicate number(s) too far from existing nodes
I don't know for sure how mkgmap should handle OSM data like this.
http://www.openstreetmap.org/way/349694725 creates a the address housenumber
915,
which also exists as a POI node
http://www.openstreetmap.org/node/2619888691
The position of the interpolated address is ~86 m east to the position of the
POI,
mkgmap r3635 tolerates up to 40 m.
I think the POI should be considered as exact, while interpolated addresses are
not.
Now, when mkgmap ignores the interpolated addresss and uses the POI, you will
see strange results for numbers close to the POI, e.g. 913 and 917 are far away
from
915 while 915 is found at the right place.
I see only one possible solution for this problem in mkgmap:
It could try to calculate the correct addr:interpolation way by adding the POI
node(s).
The result would be like what I did with
http://www.openstreetmap.org/node/3715350820
I don't know yet how complex that calculation will be.
Gerd
From: [email protected]
To: [email protected]
Date: Sat, 29 Aug 2015 00:25:17 -0300
Subject: Re: [mkgmap-dev] Commit: r3635: corrections to addr:interpolation way
checks
Hi Gerd
In the two tests
http://www.openstreetmap.org/node/3717945525#map=19/-27.24507/-52.02267
and http://www.openstreetmap.org/node/3715350820#map=19/-27.24223/-52.02223
This you changed there, in both instances in version 3635 made the
right interpolation, showing even and odd correctly.
_______________________________________________________________________________________________________
Anor C. A. de Souza
Concórdia SC
> From: [email protected]
> To: [email protected]
> Date: Fri, 28 Aug 2015 15:45:19 +0100
> Subject: [mkgmap-dev] Commit: r3635: corrections to addr:interpolation way
> checks
>
>
> Version mkgmap-r3635 was committed by gerd on Fri, 28 Aug 2015
>
> corrections to addr:interpolation way checks
>
> - The checks did not notice when a "duplicated number" belongs to a different
> cityInfo
> - The checks allowed only 5m distance between an existing address and the
> interpolated one, this is changed to 40m
> - some improvements in logging
> _______________________________________________
> mkgmap-dev mailing list
> [email protected]
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
[email protected]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
[email protected]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
[email protected]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
