February 2011 - mkgmap-dev - The mkgmap lists

Improvements to OSM Garmin maps
by Colin Smale 11 Feb '11

11 Feb '11

Hi guys, I have recently started making my own Garmin maps for my Nuvi 265WT based on OSM planet extracts. I have been downloading ready-made maps from raumbezug.de, but now I would like to improve on their maps. Their UK/Ireland maps appear to be built with --drive-on-right, which causes the odd peculiar routing instruction plus the wrong icons are shown for roundabouts. But there are some other things I would like to improve on, and I would appreciate any tips or feedback anyone may have about my ideas. Do they sound feasible? Or have better brains than mine already tried and failed? * Spoken routing instructions at motorway/trunk/etc junctions, forks and slip roads With OSM Garmin maps I often hear simple instructions like "keep right on A12" whereas the official Garmin maps manage to say something more complete like "Take exit 4, on right, onto A12 towards NewTown". Also in the official maps the single line of information about the next junction displayed at the top of the map display is not (usually) equal to the spoken "name" of the junction, whereas with the OSM maps it always seems to be equal. I am trying to find if there are "magic combinations" of tagging, mkgmap options, code and stylesheets to see if these points can be improved. Initial experiments have shown that small improvements are possible, but I am not getting anywhere near the richness of the information from the Garmin maps. * Spoken street/city names I have my unit set to UK English, but I live in the Netherlands. With the Official maps, the TTS makes a very valiant attempt at pronouncing Dutch street/city names in the Dutch way, albeit (I suspect) using English phonemes. With the OSM maps, the names are pronounced as if they were English, which is sometimes just "unclear" but very often entertaining. With the official maps, something somewhere knows it should be pronounced in Dutch. It could be a question of a header value in the IMG, it might be some kind of tag on each string to indicate the language, or it might be that each string is also stored in a phonetic form for use by TTS. Any ideas? * Speed limit icons These are not shown with the OSM maps. How can these be enabled, based on maxspeed tags? I am not initially planning to modify the java code; I would like to see how far I can get with tagging, stylesheets and options. My assumption is that TYP files only concern the way things are displayed, and are irrelevant to the above points. Colin

3 2

Base map interfering with routing
by Marko Mäkelä 11 Feb '11

11 Feb '11

When I tested routing after moving most points to resolution 24, I hit a big surprise. I let my Edge 705 calculate the route to this shop=organic: http://www.openstreetmap.org/browse/node/370814886 To my great surprise, Garmin used the base map for computing the route, no matter which mode of transportation I chose. I thought that maybe my map tile has grown too big, but splitting it into three did not help. Then, I tried routing to some nearby points, such as the bus station http://www.openstreetmap.org/browse/node/59967454, and that worked. I can navigate to that point even when using a single map tile. I wonder what is causing this? That the shop=organic could be located exactly on a way on the basemap? One more interesting point is that I did this calculation in demo mode, from about http://www.openstreetmap.org/?lat=60.3491678&lon=25.1092643&zoom=18. When the route used the base map, this position jumped to about http://www.openstreetmap.org/?lat=60.3473225&lon=25.1271307&zoom=18 which was close to the way (ref=140) on the basemap. I have successfully calculated a route to the shop before, but that was with an older basemap. The old basemap displayed a toilet symbol near my house; that's why I know. :-) Regards, Marko

5 9

mkgmap/generate-sea=extend-sea-sectors problems
by Thorsten Kukuk 11 Feb '11

11 Feb '11

Hi, I'm building my own OSM map for Garmins since quite some time, but now I wanted to fix the issue, that the sea isn't drawn correctly. For this I played with the generate-sea option. But independent to which value I set this option, I always get this java error after quite some time: java.lang.ClassCastException: java.util.TreeMap$1 incompatible with java.util.NavigableSet at uk.me.parabola.mkgmap.reader.osm.SeaGenerator.createInnerWays(SeaGenerator.java:518) at uk.me.parabola.mkgmap.reader.osm.SeaGenerator.end(SeaGenerator.java:337) at uk.me.parabola.mkgmap.reader.osm.OsmReadingHooksChain.end(OsmReadingHooksChain.java:78) at uk.me.parabola.mkgmap.reader.osm.xml.Osm5MapDataSource.load(Osm5MapDataSource.java:73) at uk.me.parabola.mkgmap.main.MapMaker.loadFromFile(MapMaker.java:145) at uk.me.parabola.mkgmap.main.MapMaker.makeMap(MapMaker.java:56) at uk.me.parabola.mkgmap.main.Main$1.call(Main.java:220) at uk.me.parabola.mkgmap.main.Main$1.call(Main.java:217) at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:314) at java.util.concurrent.FutureTask.run(FutureTask.java:149) at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:897) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:919) at java.lang.Thread.run(Thread.java:736) Exiting - if you want to carry on regardless, use the --keep-going option I tested snapshot 1812 and 1829. Any idea what causes this and how to fix/workaround this? Thanks, Thorsten

2 4

Commit: r1838: Fixed ClassCastException in SeaGenerator
by svn commit 10 Feb '11

10 Feb '11

Version 1838 was commited by wanmil on 2011-02-10 17:54:38 +0000 (Thu, 10 Feb 2011) Fixed ClassCastException in SeaGenerator Probably this happened only when using non Sun JRE or not a current JRE version but this fix prevents the ClassCastException by source coding.

1 0

Resolution on overview map.
by Pelle Svensson 10 Feb '11

10 Feb '11

Hi I can't align a detailed map correct in MapSource. The collection of maps consists of a number of small maps size around 20 by 20 km. Same type of problem: http://www.mail-archive.com/mkgmap-dev@lists.mkgmap.org.uk/msg04484.html I noticed that the resolution for level 0 and 1 is 13 resp. 12 for the overview map. This is probely the cause why the detailed map is not aligned correct relative the overview map. I tried to change the resolution by --levels="0=15, 1=13" but it seems not to be working. Is there any other way to make the overview map use finer resolution? /Pelle

1 0

crash. Too many tiles?
by Valentijn Sessink 09 Feb '11

09 Feb '11

Hi list, This should have been a bit of an experimental map, with as much tiles as possible, by having splitter do --max-nodes=150000 In the end, however, I don't get a map at all: java -enableassertions -Xmx1700m -jar ../mkgmap/dist/mkgmap.jar --make-opposite-cycleways --mapname=31000068 --latin1 --remove-short-arcs=2.8 --reduce-point-density=5.4 --lower-case --route --preserve-element-order --max-jobs --location-autofill=1 --link-pois-to-ways --description=Nederland --style=1430 --family-name="Openstreetmap 29 jan 11" --series-name=OSM-Valentijn --gmapsupp --tdbfile --net -c template.args Exception in thread "main" java.lang.IllegalArgumentException at java.nio.ByteBuffer.allocate(ByteBuffer.java:311) at uk.me.parabola.imgfmt.sys.Directory.sync(Directory.java:176) at uk.me.parabola.imgfmt.sys.ImgFS.sync(ImgFS.java:230) at uk.me.parabola.imgfmt.sys.ImgFS.close(ImgFS.java:240) at uk.me.parabola.mkgmap.combiners.GmapsuppBuilder.onFinish(GmapsuppBuilder.java:119) at uk.me.parabola.mkgmap.main.Main.endOptions(Main.java:413) at uk.me.parabola.mkgmap.CommandArgsReader.readArgs(CommandArgsReader.java:126) at uk.me.parabola.mkgmap.main.Main.main(Main.java:127) This is in the Netherlands, so nodes all over the place, and in fact, I'm getting lots and lots of "Area (51.8994140625,6.6357421875) to (52.0751953125,6.8994140625) contains 209.578 nodes but can't be split further" messages. The resulting cut has 265 tiles. A few weeks ago, I remember having made such a map - but that could just be because there were a bit fewer nodes somewhere and the number of tiles was below 256 - could just be, I don't know. (The reason for this many tiles was the idea that a Nüvi might get better bike performance on long distances if the number of data to unpack (i.e. individual tiles) would be smaller - because of the smaller tiles. Might be just a stupid idea, I don't know, I was going to check but we might never know now... ;) So I'm just passing this on, for the sake of it, don't look too long at it because it's probably not worth it; but if there's an easy fix, it is welcome ;-) Best regards, Valentijn

3 2

Re: [mkgmap-dev] [mkgmap-svn] Commit: r1826: Cities should be a lot better now.
by Steve Ratcliffe 09 Feb '11

09 Feb '11

> Version 1826 was commited by steve on 2011-02-04 23:50:16 +0000 (Fri, 04 Feb 2011) > BRANCH: index > > Cities should be a lot better now. Or maybe not... The changes were about reading the cities and fetching the names which I believe is improved although it doesn't have an affect on anything visible. I do most of my testing for street+city search on a version 6.10.2 MapSource where it has always worked. However I've now found that street+city does not work at all on a fully up-to-date version of MapSource with the exact same map. So the latest checkin may have broken city+street search again for later versions of MapSource. ..Steve

3 5

a few points
by Jozef Riha 07 Feb '11

07 Feb '11

hello, i've been playing with mkgmap and etrex legend a little bit and these are my findings: examples/styles/default/point: tourism=information {name '${name} - ${description} (${operator})'.. [0x2f0c resolution 20] - this is most likely not right as 2f0c is auto-services|rest-area-tourist-info|Restroom (and also renders as such in garmin). it could be that tourism=information couples with information=board (there's no building). hence you cannot assume that restroom are always present. amenity=place_of_worship [0x2c0b resolution 21] - this is probably correct with newer garmins but in etrex legend it renders as an attraction, probably better to replace with 0x64|0x04|manmade-places|church|Church garmin_feature_list.csv: trial -> trail i generated a map containing all possible types of point. there are quite a few that are displayed in garmin but not included in garmin_feature_list.csv. are you interested in listing them? cheers, jose

2 2

Commit: r1829: Remove POI clutter by moving most points to resolution=24.
by svn commit 07 Feb '11

07 Feb '11

Version 1829 was commited by marko on 2011-02-07 12:22:44 +0000 (Mon, 07 Feb 2011) Remove POI clutter by moving most points to resolution=24. Suggested by extremecarver.

1 0

splitter & geonames-file
by Minko 07 Feb '11

07 Feb '11

I've noticed that if I use the splitter option --geonames-file==cities5000.txt, and it can't find any cities within the splitted tile, it skips the description name. Processing mkgmap with the template.args file results in a tile with a description name of the preceding tile, because of the # in the template. On my map this results in two maps having the same description name: In the template.args file: mapname: 20010022 description: NL-Sneek input-file: 20010022.osm.gz mapname: 20010023 # description: OSM Map input-file: 20010023.osm.gz In the processed gmapsupp.img: 20010022 MPC 3742902 5 1252 20 1 20010 NL-Sneek (20010022) 20010023 MPC 324001 5 1252 20 1 20010 NL-Sneek (20010023) I suggest to leave the # out of the default description names in the template.args file, or is there another way to fix this error?

1 0