June 2012 - mkgmap-dev - The mkgmap lists

Suggestion for more flexibility in locator part of mkgmap
by Clifford Nolan 19 Sep '12

19 Sep '12

Hi Greg, Just to clarify ... the admin_levels for Ireland described here http://wiki.openstreetmap.org/wiki/Ireland/Boundaries#Boundary_Types don't seem to mention things like village boundaries - there's a big gap between admin_level=8 and admin_level=10. In situations like this, maybe the locator could be more flexible and use other kinds of boundaries too. Regards, Cliff

3 7

Problem with huge polygon - vänern sea - sweden
by toc-rox 19 Sep '12

19 Sep '12

Hi all, I'm currently working on a map of sweden. I run into a problem concerning the vänern sea (between Vänersborg and Karlstad). This is an very huge polygon / sea - 10 times larger than the german/swiss Bodensee. Only a small part of the sea will be displayed on the as water: http://gis.19327.n5.nabble.com/file/n5710048/Bildschirmfoto_2012-05-24_um_1… Questions: - Has someone a working splitter / mkgmap configuration for sweden ? - I have set the splitter option "-overlap=10000" - could this be insufficient ? Regards Klaus -- View this message in context: http://gis.19327.n5.nabble.com/Problem-with-huge-polygon-vanern-sea-sweden-… Sent from the Mkgmap Development mailing list archive at Nabble.com.

4 11

WARNING: input files have differing sort orders
by toc-rox 18 Aug '12

18 Aug '12

Hi all, what does this warning mean ? WARNING: input files have differing sort orders Regards Klaus PS: All images are compiled with mkgmap r2309 and the option "--latin1". java -Xmx6000M -jar /home/kto/Freizeitkarte-Entwicklung/tools/mkgmap/mkgmap.jar --max-jobs=3 --license-file=license.txt --index --gmapsupp --product-id=1 --family-id=5819 --family-name="Freizeitkarte_Nordrhein-Westfalen 12.6" --series-name="Freizeitkarte_Nordrhein-Westfalen 12.6" --description="Freizeitkarte_Nordrhein-Westfalen 12.6" --overview-mapnumber=58190000 --product-version=1206 5819*.img 5819.TYP Time started: Fri Jun 29 14:13:25 UTC 2012 WARNING: input files have differing sort orders Time finished: Fri Jun 29 14:13:40 UTC 2012 -- View this message in context: http://gis.19327.n5.nabble.com/WARNING-input-files-have-differing-sort-orde… Sent from the Mkgmap Development mailing list archive at Nabble.com.

5 11

Address search issues
by Ralf kleineisel 29 Jul '12

29 Jul '12

Hi, I'm having a hard time getting the address search to work. I have split the area of Germany from the planet pbf file. Then I build the Garmin IMGs with this commandline: /usr/java/jdk1.6.0_24/bin/java -enableassertions -Xmx2048m -jar /data/gps/mkgmap-r2294/mkgmap.jar and this mkgmap config file: ### description=OSM_Germany_Alps country-name=Germany country-abbr=DE region-name=Germany region-abbr=DE name-tag-list="name:mkgmap,name:de,int_name,name:en,name" overview-mapname=OSM_Germany_Alps overview-mapnumber=10000000 latin1 reduce-point-density=4 merge-lines=yes family-id=1331 product-id=1 series-name=OSM_SRTM_Germany_Alps family-name=OSM_Germany area-name=Germany index style-file=/data/gps/mkgmap-style-toponew_test120611/ tdbfile=no route=yes road-name-pois=0x640a add-pois-to-areas make-opposite-cycleways=yes remove-short-arcs adjust-turn-headings link-pois-to-ways bounds=/data/gps/boundary/bounds/ location-autofill=bounds,is_in,nearest generate-sea=extend-sea-sectors,multipolygon,close-gaps=1000,floodblocker,fbgap=200,fbthres=20,fbratio=0.5 mapname: 10000001 description: FR-Besancon input-file: 10000001.osm.gz ... [more mapname/description/input-file lines] ### The directory /data/gps/boundary/bounds/ contains the unzipped file http://www.navmaps.eu/wanmil/bounds_20120401.zip The "lines", "points" and "polygons" files in my style contain the top lines from the default style which set the location tags AFAIK. I make the mapsource files afterwards: ### description=OSM_SRTM_Germany country-name=Germany country-abbr=DE region-name=Germany region-abbr=DE name-tag-list="name:mkgmap,name:de,int_name,name:en,name" overview-mapname=OSM_SRTM_Germany overview-mapnumber=10000000 latin1 series-name=OSM_SRTM_Germany area-name=Test family-id=1331 product-id=1 family-name=OSM_Germany index tdbfile=yes input-file: 10000001.img ... [more tile files] ### When I try the address search in mapsource I get this: City search for "Berlin": Berlin is found, but mapsource displays "Berlin, NOT_SET, DEU". When I search for a street in Berlin (say, Alexanderstrasse, I can see the POI on the map) I cannot enter "Berlin" in the City field. As soon as I enter the "i" mapsource deletes it again. When I enter the postcode (10178 for Alexanderstrasse, Berlin) the street is found. Mapsource displays "Alexanderstrasse, Not_Set, NOT_SET, DEU". I tried this for various cities and streets. When I build a gmapsupp.img it behaves rather the same in my eTrex Legend HCx. Any ideas? Is there anything wrong with the bounds data or the style files?

3 10

mkgmap crashes if nodes are missing
by NopMap 29 Jul '12

29 Jul '12

I have been debugging a crash of mkgamp where it fails with the exception java.lang.IndexOutOfBoundsException: Index: 0, Size: 0 at java.util.ArrayList.RangeCheck(Unknown Source) at java.util.ArrayList.get(Unknown Source) at uk.me.parabola.mkgmap.reader.osm.MultiPolygonRelation$JoinedWay.<init >(MultiPolygonRelation.java:2199) at uk.me.parabola.mkgmap.reader.osm.MultiPolygonRelation.joinWays(MultiP olygonRelation.java:245) at uk.me.parabola.mkgmap.reader.osm.MultiPolygonRelation.processElements (MultiPolygonRelation.java:763) at uk.me.parabola.mkgmap.reader.osm.ElementSaver.addRelation(ElementSave r.java:166) at uk.me.parabola.mkgmap.reader.osm.xml.Osm5XmlHandler$SaxHandler.endEle ment(Osm5XmlHandler.java:182) The reason seems to be inconsistent data: a way without the nodes it references. As mkgmap siletntly drops missing ways, but crashes on missing nodes I assume this is a bug. This can be fixed by changing Osm5XmlHandler, line 263 to if ("way".equals(type)){ el = saver.getWay(id); // remove ways without points if( el != null && ((Way)el).getPoints().isEmpty() ) el = null; } Maybe someone would like to apply this fix to make mkgmap more robust. bye Nop -- View this message in context: http://gis.19327.n5.nabble.com/mkgmap-crashes-if-nodes-are-missing-tp571181… Sent from the Mkgmap Development mailing list archive at Nabble.com.

2 5

[PATCH v1] Precompiled sea generator
by WanMil 26 Jun '12

26 Jun '12

Hi, attached is the generator for precompiled sea data. Additionally you need some geotools libs (they are not downloaded by ivy). You can download it from http://sourceforge.net/projects/geotools/files/ The required libs are: commons-pool-1.5.4.jar gt-api-2.7.5.jar gt-cql-2.7.5.jar gt-data-2.7.5.jar gt-main-2.7.5.jar gt-metadata-2.7.5.jar gt-opengis-2.7.5.jar gt-referencing-2.7.5.jar gt-shapefile-2.7.5.jar jsr-275-1.0-beta-2.jar jts-1.11.jar vecmath-1.3.2.jar The generator can be started with java -cp <mkgmap classes dir> uk.me.parabola.mkgmap.sea.optional.PrecompSeaGenerator <shapefilename> <projection> <outputdir> shapefilename: the filename of the .shp file projection: WGS84 or any EPSG:xxxx outputdir: the directory to store the precompiled files The generator is not included in the mkgmap.jar file because the geotools libs are not automatically downloaded by ivy. If anyone can give an ivy dependency for the libs it is possible to change that. There are some performance settings that might be changed on your computer to better meet the memory/CPU settings of your computer. Have a look at the PrecompSeaGenerator constructor. The defined settings are ok for 3gb memory (-Xmx3g). WanMil

2 6

Re: [mkgmap-dev] Province names
by Nils Leandersson 25 Jun '12

25 Jun '12

I got it to work. I added the name tag list option to my mkgmap call (--location-autofill=bounds --name-tag-list=int_name,name:en,name) in addition to the name tag list in the style sheet options file. Theory: It appears that the name tag list in the options style sheet has no influence on the POI's addresses calculated from preprocessed boundaries. ----- Original Message ----- From: Nils Leandersson To: Development list for mkgmap Sent: Monday, June 25, 2012 5:59 PM Subject: Re: [mkgmap-dev] Province names Thanks. I am referring to the province name added to all POI's address information so neither 1 or 2. Hopefully this screen shot from my Zumo explains it better: This is the listing I get when browsing cities. The three first ones are okay because they are located within provinces that have English names in the 'name' tag in the relation outlining the province. The last, Bangkok (city), is within Bangkok province. The relation for Bangkok province is properly tagged with "กรุงเทพมหานคร" in it's 'name' tag and "Bangkok" in it's 'name:en' tag. As you can see, the 'name:en' is not used but rather Thai script "Latinized". Bangkok's relation: http://ra.osmsurround.org/analyzeRelation?relationId=92277&_noCache=on# Samut Sakhon's relation: http://ra.osmsurround.org/analyzeRelation?relationId=1908816&_noCache=on# ----- Original Message ----- From: "WanMil" <wmgcnfg(a)web.de> To: <garvan.okeeffe.esssuap(a)moeys.gov.kh>; "Development list for mkgmap" <mkgmap-dev(a)lists.mkgmap.org.uk> Sent: Monday, June 25, 2012 4:07 PM Subject: Re: [mkgmap-dev] Province names > Nils, > > after reading Garvans comment I wonder if I understood your problem > correctly. > > What do you mean that the province name uses that name tag instead of > name:en? > 1. All POIs within the province use the name tag for the province name? > 2. Only the boundary itself is displayed with the name tag instead of > name:en? > 3. both 1. and 2. > > If the answer is 1, then I have to try to reproduce the problem. Garvans > comment should not be relvant to 1. > > If the answer is 2, please follow Garvans hints. > > WanMil > >> About a year ago I reported that the name tag list was not working the >> way I expected. I think it was not applied to relations, but I made some >> modifications to my style file to work around the issue. I am not sure >> if this is related to the problems you are having now, but I looked up >> the old thread for you to check out. This is the reply to my observations. >> >> http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2010q4/009593.html >> >> Best regards, >> >> Garvan >> >> >> On 24-Jun-12 13:49, Nils Leandersson wrote: >>> <snip> >>> >>> The final result is the same, the "name" tag is used rather than the >>> "name:en" tag. >>> >>> Can anyone confirm they can use any other name tag than the default 'name', >>> and if so, what options was used for osmosis and mkgamp? >>> >>> >>> >> >> _______________________________________________ >> mkgmap-dev mailing list >> mkgmap-dev(a)lists.mkgmap.org.uk >> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev > > _______________________________________________ > mkgmap-dev mailing list > mkgmap-dev(a)lists.mkgmap.org.uk > http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

1 0

Province names
by Nils Leandersson 25 Jun '12

25 Jun '12

I am trying to make a map that uses the English names for provinces in SE Asia. The province shown on my Zumo for POIs in Bangkok is a "Latinized" กรุงเทพมหานคร rather than "Bangkok" from the name:en tag. The relation in question (92277/6) is tagged name=กรุงเทพมหานคร and name:en=Bangkok. I use the --latin option when I compile the map. I extract boundaries with: bin\Osmosis.bat --rb asia.osm.pbf --tf accept-ways boundary=administrative --tf accept-relations boundary=administrative --used-node --write-pbf boundaries.osm.pbf omitmetadata=true I compile boundaries with: java -ea -Xmx1000m -jar mkgmap/mkgmap.jar --createboundsfile=boundaries.osm.pbf --location-autofill=bounds name-tag-list=name:en,name What am I doing wrong? Do the tags on the boundaries play a role or are only the tags on the relations used? What would be the way to check all admin4 boundaries' tags in SE Asia? The boundaries seem to be tagged inconsistent.

3 6

Issue when routing across two countries
by Enrico Liboni 25 Jun '12

25 Jun '12

Routing works excellent on my Nuvi except when the Start/End points are in different countries; if it is the case: 1. the device takes a lot (minutes) to calculate the route - also if the two points/cities are pretty close, while for routes within the same country takes seconds. 2. the route to follow (purple) is mostly translated from the real streets, a picture <http://imageshack.us/photo/my-images/401/garminrouteissue.jpg/>( http://imageshack.us/photo/my-images/401/garminrouteissue.jpg/ - sorry for the quality) better explains the behavior: the red street is the highway, that's the only possible way. The purple line is the displayed route, as you can see they are not matching. This happens only when start/end points are in different countries. Picture is taken routing from Bolzano, Italy to Innsbruck, Austria. I create gmapsupp from geofabix's pbf files, I tried today with mkgmap version r2306, but I experienced the issue as well with older versions I tried with. I use standard command lines option only "--route --remove-short-arcs --add-pois-to-areas --index --gmapsupp". Bounds are from navmaps.eu Any clue? Thanks! Enrico

6 8

Commit: r2309: Reverse the header change that was made in the dev-index branch.
by svn commit 24 Jun '12

24 Jun '12

Version 2309 was commited by steve on 2012-06-24 19:33:53 +0100 (Sun, 24 Jun 2012) Reverse the header change that was made in the dev-index branch.

1 0