City search problem

classic Classic list List threaded Threaded
16 messages Options
Reply | Threaded
Open this post in threaded view
|

City search problem

ValentinAK
Hello.
I release daily assemblies of all countries that were part of the ex-USSR
and several other countries ( maptourist.org <https://maptourist.org>  ).
The amount of data becomes very large and will soon reach 4GB of the maximum
size of one gmapsupp container. Therefore, I decided to divide my assembly
into two parts - Eastern and Western. I made the separation along the border
(admin_level = 3). But I found a problem when searching for cities. Now I
will try to explain what I discovered:

I prepared two parts of the OSM data with a common border and compiled
east-gmapsupp.img and west-gmapsupp.img. I use Splitter v591 & Mkgmap v4240.
Splitter processing the western part of the data makes several rectangular
areas. The data clipping boundary crossing several tiles. As a result, half
the area of ​​these tiles contains no data.
The empty space of these tiles covers a non-empty map from the eastern part.
Now I activate east-gmapsupp.img and west-gmapsupp.img in my Garmin
Nuvi-1490 and on the empty space of the described western tiles, I get a
problem when searching for all eastern cities in this area of ​​overlap.
Only the search for cities does not work. Address search and routing work
fine. I tried the same and different FID. The problem is still here.

I plan to compile more maps of many other countries, but this problem makes
it difficult to use two adjacent maps simultaneously.
What solution?

Links for try this on my fresh files:
https://maptourist.org/osm-garmin/gmapsupp-east-EN.zip
<https://maptourist.org/osm-garmin/gmapsupp-east-EN.zip>  
https://maptourist.org/osm-garmin/gmapsupp-west-EN.zip
<https://maptourist.org/osm-garmin/gmapsupp-west-EN.zip>  

Best regards
Valentin

<http://gis.19327.n8.nabble.com/file/t318348/City-search-problem.gif>




--
Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

Gerd Petermann
Hi Valentin,

no idea if we can solve that problem in mkgmap, sounds more like a problem in the Garmin software.
I suggest to use a different way to split the maps so that you don't have empty areas. In other words, don't use the admin boundary but a
vertical line to split the OSM file. I think you can use splitter for that with the polygon-desc-file option.

Gerd


________________________________________
Von: mkgmap-dev <[hidden email]> im Auftrag von ValentinAK <[hidden email]>
Gesendet: Donnerstag, 4. Oktober 2018 01:58
An: [hidden email]
Betreff: [mkgmap-dev] City search problem

Hello.
I release daily assemblies of all countries that were part of the ex-USSR
and several other countries ( maptourist.org <https://maptourist.org>  ).
The amount of data becomes very large and will soon reach 4GB of the maximum
size of one gmapsupp container. Therefore, I decided to divide my assembly
into two parts - Eastern and Western. I made the separation along the border
(admin_level = 3). But I found a problem when searching for cities. Now I
will try to explain what I discovered:

I prepared two parts of the OSM data with a common border and compiled
east-gmapsupp.img and west-gmapsupp.img. I use Splitter v591 & Mkgmap v4240.
Splitter processing the western part of the data makes several rectangular
areas. The data clipping boundary crossing several tiles. As a result, half
the area of ​​these tiles contains no data.
The empty space of these tiles covers a non-empty map from the eastern part.
Now I activate east-gmapsupp.img and west-gmapsupp.img in my Garmin
Nuvi-1490 and on the empty space of the described western tiles, I get a
problem when searching for all eastern cities in this area of ​​overlap.
Only the search for cities does not work. Address search and routing work
fine. I tried the same and different FID. The problem is still here.

I plan to compile more maps of many other countries, but this problem makes
it difficult to use two adjacent maps simultaneously.
What solution?

Links for try this on my fresh files:
https://maptourist.org/osm-garmin/gmapsupp-east-EN.zip
<https://maptourist.org/osm-garmin/gmapsupp-east-EN.zip>
https://maptourist.org/osm-garmin/gmapsupp-west-EN.zip
<https://maptourist.org/osm-garmin/gmapsupp-west-EN.zip>

Best regards
Valentin

<http://gis.19327.n8.nabble.com/file/t318348/City-search-problem.gif>




--
Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

popej
Hi Gerd,

Probably the solution is to use background polygon trimmed by
country/region area. I think GPS look for background at current position
to determine which map to use for a search. Some GPS allow to select
map, but mostly it goes automatically, GPS selects a map and then
doesn't look for another, even if there is no hits. So near the border,
where tiles overlap, GPS can select a tile from a wrong map.

I hope that trimmed background could make a clean division at the
border, even if tiles coordinates overlap.

--
Best regards,
Andrzej
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

ValentinAK
In reply to this post by Gerd Petermann
Hi Gerd.
Thanks for the answer!
Vertical line to split is a good solution in case with my ex-USSR map only.
But not usable for another countries.

Best regards,
Valentin



--
Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

ValentinAK
In reply to this post by popej
Hi Andrzej.
Earlier, I also wanted to use the --polygon-file Splitter option, but found
it unusable. Only 40 vertices are allowed in the poly file. This is
extremely small for repeating the border for a background polygon.
Why this restriction? Very lacking the ability of unlimited poly file!

Best regards,
Valentin



--
Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

ValentinAK
I received a user review for my ex-USSR map on his Nuvi-2797LM. There is
another problem with overlapping eastern and western parts. In the area with
the problem as I described earlier, Garmin shows nothing on this device. :(

<http://gis.19327.n8.nabble.com/file/t318348/39097.png>

<http://gis.19327.n8.nabble.com/file/t318348/34128.png>



--
Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

Gerd Petermann
In reply to this post by ValentinAK
Hi Valentin,

there is no such limit. The polygon can be very complex. Splitter just uses a different algo if this happens. The problem here is that splitter tries to find rectangles that fit into the polygon AND don't have more than the given nodes AND are not too empty AND
they have to be on the selected raster. The more complex the polygon the less likely it is to find a solution for this.
You can change the limit in the source and try what happens.

Gerd

________________________________________
Von: mkgmap-dev <[hidden email]> im Auftrag von ValentinAK <[hidden email]>
Gesendet: Donnerstag, 4. Oktober 2018 14:51
An: [hidden email]
Betreff: Re: [mkgmap-dev] City search problem

Hi Andrzej.
Earlier, I also wanted to use the --polygon-file Splitter option, but found
it unusable. Only 40 vertices are allowed in the poly file. This is
extremely small for repeating the border for a background polygon.
Why this restriction? Very lacking the ability of unlimited poly file!

Best regards,
Valentin



--
Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

ValentinAK
Hi Gerd,
I appreciate you for this explanation.
Ok, I understand now.
Maybe there is a reason to add another new processing algorithm in Mkgmap?
How I see it:
1. The Splitter is doing its job as it is now. Without any changes.
2. Mkgmap takes a specific the boundary relation or admin_level from the
data or a poly file for the final generation of the background polygon.

This feature in Mkgmap is really lacking!
I do not know how to do it because I am not a programmer. Just an idea.

Best regards,
Valentin



--
Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

Gerd Petermann
Hi all,

reg. background polygon:
I wonder how this is supposed to work. I see these possibilities:
1) Try to calculate a (multi)polygon using the admin_level=x ways and cut all lines and shapes exactly along the boundaries of the polygon
2) Like 1) but increase the size of the polygon by a given percentage (JOSM has such a function called "Make parallel copies of ways").
3) Let the user specify a (multi)polygon and cut all lines and shapes exactly along the boundaries of the polygon. The user can decide if he wants overlaps or not.

I see one big problem with 1) and 2) : What should happen when the boundaries don't form a polygon? I think we end up with the same problem as in the sea generator, therefore I think 3) is the way to go.

Comments?
Gerd

________________________________________
Von: mkgmap-dev <[hidden email]> im Auftrag von ValentinAK <[hidden email]>
Gesendet: Donnerstag, 4. Oktober 2018 19:42
An: [hidden email]
Betreff: Re: [mkgmap-dev] City search problem

Hi Gerd,
I appreciate you for this explanation.
Ok, I understand now.
Maybe there is a reason to add another new processing algorithm in Mkgmap?
How I see it:
1. The Splitter is doing its job as it is now. Without any changes.
2. Mkgmap takes a specific the boundary relation or admin_level from the
data or a poly file for the final generation of the background polygon.

This feature in Mkgmap is really lacking!
I do not know how to do it because I am not a programmer. Just an idea.

Best regards,
Valentin



--
Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

popej
Hi Gerd,

I'm not sure if there should be any cutting of objects. My guess is,
that only background polygon is essential. Maybe would be enough to
simply discard objects, which are fully outside of background?

All this needs some testing.

--
Best regards,
Andrzej
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

ValentinAK
Hi.
Background rectangle polygons with no data are make a different behavior in
various Garmin devices. They are all not good usually.
I like the third solution method that Gerd suggested.

Best regards,
Valentin



--
Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

Gerd Petermann
In reply to this post by popej
Hi Andrzej,

sorry for the late response. Maybe I got you wrong here. My understanding is that cutting with a polygon is a complex task and I think we don't
have code for that in mkgmap. On the other hand, without cutting we might already solve the search problem. Was that your idea?

Gerd

________________________________________
Von: mkgmap-dev <[hidden email]> im Auftrag von Andrzej Popowski <[hidden email]>
Gesendet: Freitag, 5. Oktober 2018 18:37
An: [hidden email]
Betreff: Re: [mkgmap-dev] City search problem

Hi Gerd,

I'm not sure if there should be any cutting of objects. My guess is,
that only background polygon is essential. Maybe would be enough to
simply discard objects, which are fully outside of background?

All this needs some testing.

--
Best regards,
Andrzej
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

popej
Hi Gerd,

yes, I think that we could use irregular background without trimming
other objects. At least for some testing.

I'm not sure, what would be the easiest way to implement it. Maybe add
background polygon as explicit data in input source and include
processing for it in the style?

--
Best regards,
Andrzej
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

Gerd Petermann
Hi Andrzej,

I'd prefer to reuse the code that is used with --dem-polygon to specify the polygon itself.
Maybe you can play with mp format? If mkgmap finds a background polygon in the mp file it will not add one.

My own motivation to start coding is still rather small, maybe because I think of it as a political problem, maybe because of the nice wether here in Germany and my ongoing problems with a rather expensive new bicycle ;-)

Gerd

________________________________________
Von: mkgmap-dev <[hidden email]> im Auftrag von Andrzej Popowski <[hidden email]>
Gesendet: Dienstag, 16. Oktober 2018 12:16
An: [hidden email]
Betreff: Re: [mkgmap-dev] City search problem

Hi Gerd,

yes, I think that we could use irregular background without trimming
other objects. At least for some testing.

I'm not sure, what would be the easiest way to implement it. Maybe add
background polygon as explicit data in input source and include
processing for it in the style?

--
Best regards,
Andrzej
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

popej
Hi Gerd,

I'll try to do some tests with mp, when I find some time. I wasn't
aware, that mkgmap accept background with this format.

Does mkgmap process statement "background=Y"? I wonder if background
object in img has any distinctive attributes.

--
Best regards,
Andrzej

_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Reply | Threaded
Open this post in threaded view
|

Re: City search problem

Gerd Petermann
Hi Andrzej,

I have no idea about the details. I just know these lines in mkgmap source PolishMapDataSource.java:
...
                if (addBackground && !havePolygon4B)
                        addBackground();
...
                        if(type == 0x4b)
                                havePolygon4B = true;
...

________________________________________
Von: mkgmap-dev <[hidden email]> im Auftrag von Andrzej Popowski <[hidden email]>
Gesendet: Mittwoch, 17. Oktober 2018 13:13
An: [hidden email]
Betreff: Re: [mkgmap-dev] City search problem

Hi Gerd,

I'll try to do some tests with mp, when I find some time. I wasn't
aware, that mkgmap accept background with this format.

Does mkgmap process statement "background=Y"? I wonder if background
object in img has any distinctive attributes.

--
Best regards,
Andrzej

_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________
mkgmap-dev mailing list
[hidden email]
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev