Re: [mkgmap-dev] splitter r246

Uli, can you please retry and give the exact error message of mkgmap?
I guess that one of your tile contains a large polygon that is now contained completely in the tile due to keep-complete. Maybe a part of mkgmap has a problem with this large polygon (e.g. subdivision handling). But this is wild guessing unless you can provide the exact error message. Wee need all information about your steps to create the germany+iceland.osm.pbf and your mkgmap parameters and style file so that we can try to reproduce the error.
Thanks WanMil
Hi WanMil, i described the way to produce the germany+iceland.osm.pbf in this post: http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2012q4/015462.html I tried again using dataset from today with following parameters: Splitter call: start /low /b /wait java -Xmx1348m -jar ..\bin\splitter.jar --max-threads=2 --keep-complete --overlap=0 --output=pbf --max-areas=256 --mapid=66%FID%001 --max-nodes=1200000 ..\OSM-Data\%osmfile% mkgmap call: start /low /b /wait java -Xmx1348m -jar ..\bin\mkgmap.jar --max-jobs=3 --generate-sea=no-sea-sectors,extend-sea-sectors --country-name="%country%" --family-id=%FID% --mapname=66%FID%001 --overview-mapname=66%FID%000 --series-name="OSM-%country%" --family-name="OpenStreetMap: %country%" --style-file=..\bin\styles\ --style=%style% --description="%country%" -c ..\bin\template_basic.args -c template.args %TypFileToAdd% template_basic.args: gmapsupp region-name:OSM latin1 product-id:1 area-name:OSM tdbfile route index bounds:..\bin\bounds.zip location-autofill:bounds,is_in,nearest coastlinefile=..\bin\coastlines_europe.osm.pbf delete-tags-file:..\bin\Tags_to_delete reduce-point-density-polygon:5 reduce-point-density:4 merge-lines min-size-polygon:8 remove-short-arcs:3 adjust-turn-headings link-pois-to-ways code-page:1252 Tags_to_delete: highway=proposed The mkgmpap message is as follows: There is not enough room in a single garmin map for all the input data The .osm file should be split into smaller pieces first. That's all, no further details and mkgmap continues with the other tiles. The defective tile with size 0 is 66923055.osm.pbf Regards, Uli

Am 26.11.2012 12:29, schrieb UliBaer:
Uli, can you please retry and give the exact error message of mkgmap?
I guess that one of your tile contains a large polygon that is now contained completely in the tile due to keep-complete. Maybe a part of mkgmap has a problem with this large polygon (e.g. subdivision handling). But this is wild guessing unless you can provide the exact error message. Wee need all information about your steps to create the germany+iceland.osm.pbf and your mkgmap parameters and style file so that we can try to reproduce the error.
Thanks WanMil
Hi WanMil, i described the way to produce the germany+iceland.osm.pbf in this post: http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2012q4/015462.html
I tried again using dataset from today with following parameters:
Splitter call: start /low /b /wait java -Xmx1348m -jar ..\bin\splitter.jar --max-threads=2 --keep-complete --overlap=0 --output=pbf --max-areas=256 --mapid=66%FID%001 --max-nodes=1200000 ..\OSM-Data\%osmfile%
mkgmap call: start /low /b /wait java -Xmx1348m -jar ..\bin\mkgmap.jar --max-jobs=3 --generate-sea=no-sea-sectors,extend-sea-sectors --country-name="%country%" --family-id=%FID% --mapname=66%FID%001 --overview-mapname=66%FID%000 --series-name="OSM-%country%" --family-name="OpenStreetMap: %country%" --style-file=..\bin\styles\ --style=%style% --description="%country%" -c ..\bin\template_basic.args -c template.args %TypFileToAdd%
template_basic.args: gmapsupp region-name:OSM latin1 product-id:1 area-name:OSM tdbfile route index bounds:..\bin\bounds.zip location-autofill:bounds,is_in,nearest coastlinefile=..\bin\coastlines_europe.osm.pbf delete-tags-file:..\bin\Tags_to_delete reduce-point-density-polygon:5 reduce-point-density:4 merge-lines min-size-polygon:8 remove-short-arcs:3 adjust-turn-headings link-pois-to-ways code-page:1252
Tags_to_delete: highway=proposed
The mkgmpap message is as follows: There is not enough room in a single garmin map for all the input data The .osm file should be split into smaller pieces first.
That's all, no further details and mkgmap continues with the other tiles. The defective tile with size 0 is 66923055.osm.pbf
Regards, Uli
Thanks! Can you upload the defective tile 66923055.osm.pbf to http://files.mkgmap.org.uk/ ? That makes it easier for me to start :-) WanMil

Am 26.11.2012 12:29, schrieb UliBaer:
Uli, can you please retry and give the exact error message of mkgmap?
I guess that one of your tile contains a large polygon that is now contained completely in the tile due to keep-complete. Maybe a part of mkgmap has a problem with this large polygon (e.g. subdivision handling). But this is wild guessing unless you can provide the exact error message. Wee need all information about your steps to create the germany+iceland.osm.pbf and your mkgmap parameters and style file so that we can try to reproduce the error.
Thanks WanMil
Hi WanMil, i described the way to produce the germany+iceland.osm.pbf in this post: http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2012q4/015462.html
I tried again using dataset from today with following parameters:
Splitter call: start /low /b /wait java -Xmx1348m -jar ..\bin\splitter.jar --max-threads=2 --keep-complete --overlap=0 --output=pbf --max-areas=256 --mapid=66%FID%001 --max-nodes=1200000 ..\OSM-Data\%osmfile%
mkgmap call: start /low /b /wait java -Xmx1348m -jar ..\bin\mkgmap.jar --max-jobs=3 --generate-sea=no-sea-sectors,extend-sea-sectors --country-name="%country%" --family-id=%FID% --mapname=66%FID%001 --overview-mapname=66%FID%000 --series-name="OSM-%country%" --family-name="OpenStreetMap: %country%" --style-file=..\bin\styles\ --style=%style% --description="%country%" -c ..\bin\template_basic.args -c template.args %TypFileToAdd%
template_basic.args: gmapsupp region-name:OSM latin1 product-id:1 area-name:OSM tdbfile route index bounds:..\bin\bounds.zip location-autofill:bounds,is_in,nearest coastlinefile=..\bin\coastlines_europe.osm.pbf delete-tags-file:..\bin\Tags_to_delete reduce-point-density-polygon:5 reduce-point-density:4 merge-lines min-size-polygon:8 remove-short-arcs:3 adjust-turn-headings link-pois-to-ways code-page:1252
Tags_to_delete: highway=proposed
The mkgmpap message is as follows: There is not enough room in a single garmin map for all the input data The .osm file should be split into smaller pieces first.
That's all, no further details and mkgmap continues with the other tiles. The defective tile with size 0 is 66923055.osm.pbf
Regards, Uli
Thanks! Can you upload the defective tile 66923055.osm.pbf to http://files.mkgmap.org.uk/ ? That makes it easier for me to start :-)
WanMil
Hi Uli, the reason for the error is the new split algorithm introduced with splitter r245 and r246. But the split algorithm is *not* defective. It's more an unfortunate combination of circumstances. The new split algorithm created the tile 66923055 which covers most of iceland and a very small part of germany. You are using an europe coastline file which means that the whole coastline of this tile should be generated. But this is not iceland and germany only but also big parts of scotland, ireland, england and the west coast of norway. This big amount of data exceeds the tile limits which causes the given error message. The split algorithm was working correctly by putting not more than the configured number of nodes into the tile: 1200000. (I did not check that but I assume this is correct.) But this does not and cannot consider the coastline part blanked out in your input data (iceland+germany). So with your "two part input data" you have to edit your areas.list file to avoid such situation. Some small hints about your mkgmap configuration: * I propose to use precompiled sea files (e.g. http://www.navmaps.eu/wanmil/sea_20121020.zip). It's faster and takes less memory. * Since r2335 location-autofill=bounds is silently ignored. So you might change your template_basic.args to bounds:..\bin\bounds.zip location-autofill:is_in,nearest * AFAIK latin1 and code-page:1252 are the same. So you can remove one of it. Have fun! WanMil

the reason for the error is the new split algorithm introduced with splitter r245 and r246. But the split algorithm is *not* defective. It's more an unfortunate combination of circumstances.
The new split algorithm created the tile 66923055 which covers most of iceland and a very small part of germany. You are using an europe coastline file which means that the whole coastline of this tile should be generated. But this is not iceland and germany only but also big parts of scotland, ireland, england and the west coast of norway. This big amount of data exceeds the tile limits which causes the given error message. The split algorithm was working correctly by putting not more than the configured number of nodes into the tile: 1200000. (I did not check that but I assume this is correct.) But this does not and cannot consider the coastline part blanked out in your input data (iceland+germany). So with your "two part input data" you have to edit your areas.list file to avoid such situation.
I think the solution would be to use the new polygon-file parameter. If splitter would not generate tiles which have large parts outside of the polygon (this is on my todo list now) it would be unlikely that the coastline data adds many points. Do you agree? Gerd

the reason for the error is the new split algorithm introduced with splitter r245 and r246. But the split algorithm is *not* defective. It's more an unfortunate combination of circumstances.
The new split algorithm created the tile 66923055 which covers most of iceland and a very small part of germany. You are using an europe coastline file which means that the whole coastline of this tile should be generated. But this is not iceland and germany only but also big parts of scotland, ireland, england and the west coast of norway. This big amount of data exceeds the tile limits which causes the given error message. The split algorithm was working correctly by putting not more than the configured number of nodes into the tile: 1200000. (I did not check that but I assume this is correct.) But this does not and cannot consider the coastline part blanked out in your input data (iceland+germany). So with your "two part input data" you have to edit your areas.list file to avoid such situation.
I think the solution would be to use the new polygon-file parameter. If splitter would not generate tiles which have large parts outside of the polygon (this is on my todo list now) it would be unlikely that the coastline data adds many points.
Do you agree?
Gerd
Yes. I assume it is possible to define multiple separated polygons within the polygon file? WanMil

WanMil wrote
Yes. I assume it is possible to define multiple separated polygons within the polygon file?
Yes, it is. Gerd -- View this message in context: http://gis.19327.n5.nabble.com/splitter-r246-tp5737445p5737824.html Sent from the Mkgmap Development mailing list archive at Nabble.com.

Am 26.11.2012 23:13, schrieb WanMil:
Some small hints about your mkgmap configuration: * I propose to use precompiled sea files (e.g. http://www.navmaps.eu/wanmil/sea_20121020.zip). It's faster and takes less memory.
Hi WanMil, is it possible to generate the precompiled files by myself? I haven't found anything in the wiki how to do this. I remember that I need a special tool, which have to be compiled. Would it be possible to publish a binary? If libraries are problematic, they could be only listed in a txt-file and the user downloads this separate from other sources. Is it possible to use the shape-files from http://openstreetmapdata.com/data ? Henning

Am 26.11.2012 23:13, schrieb WanMil:
Some small hints about your mkgmap configuration: * I propose to use precompiled sea files (e.g. http://www.navmaps.eu/wanmil/sea_20121020.zip). It's faster and takes less memory.
Hi WanMil, is it possible to generate the precompiled files by myself? I haven't found anything in the wiki how to do this. I remember that I need a special tool, which have to be compiled. Would it be possible to publish a binary? If libraries are problematic, they could be only listed in a txt-file and the user downloads this separate from other sources.
Is it possible to use the shape-files from http://openstreetmapdata.com/data ?
Henning
Hi Henning, there is a readme how to precompile the sea files: http://www.mkgmap.org.uk/websvn/filedetails.php?repname=mkgmap&path=%2Ftrunk... It would be possible to publish a binary. Maybe we can integrate into the existing ant scripts if you provide it. I think the libs are not problematic. These are some part of the geotools lib. You can use the shape files from openstreetmapdata. I am using them myself. Beware that you need the land polygons and not the sea polygons. It's better to use the split ones because the others are memory exhaustive. WanMil

Am 27.11.2012 13:51, schrieb WanMil:
Am 26.11.2012 23:13, schrieb WanMil:
Some small hints about your mkgmap configuration: * I propose to use precompiled sea files (e.g. http://www.navmaps.eu/wanmil/sea_20121020.zip). It's faster and takes less memory. Hi WanMil, is it possible to generate the precompiled files by myself? I haven't found anything in the wiki how to do this. I remember that I need a special tool, which have to be compiled. Would it be possible to publish a binary? If libraries are problematic, they could be only listed in a txt-file and the user downloads this separate from other sources.
Is it possible to use the shape-files from http://openstreetmapdata.com/data ?
Henning
Hi Henning,
there is a readme how to precompile the sea files: http://www.mkgmap.org.uk/websvn/filedetails.php?repname=mkgmap&path=%2Ftrunk... I've added it to: http://wiki.openstreetmap.org/wiki/Mkgmap/help/options#Using_precompiled_sea Maybe someone with more java-knowledge could explain the start of PrecompSeaGenerator a little bit more.
It would be possible to publish a binary. That would be very nice, but not highest priority ;) It shouldn't be a problem to download Geotools libary separate.
Henning

WanMil wrote
Hi Uli,
the reason for the error is the new split algorithm introduced with splitter r245 and r246. But the split algorithm is *not* defective. It's more an unfortunate combination of circumstances.
The new split algorithm created the tile 66923055 which covers most of iceland and a very small part of germany. You are using an europe coastline file which means that the whole coastline of this tile should be generated. But this is not iceland and germany only but also big parts of scotland, ireland, england and the west coast of norway. This big amount of data exceeds the tile limits which causes the given error message. The split algorithm was working correctly by putting not more than the configured number of nodes into the tile: 1200000. (I did not check that but I assume this is correct.) But this does not and cannot consider the coastline part blanked out in your input data (iceland+germany). So with your "two part input data" you have to edit your areas.list file to avoid such situation.
Some small hints about your mkgmap configuration: * I propose to use precompiled sea files (e.g. http://www.navmaps.eu/wanmil/sea_20121020.zip). It's faster and takes less memory.
* Since r2335 location-autofill=bounds is silently ignored. So you might change your template_basic.args to bounds:..\bin\bounds.zip location-autofill:is_in,nearest
* AFAIK latin1 and code-page:1252 are the same. So you can remove one of it.
Have fun! WanMil
Hi WanMil, thanks for your useful tips, i'll incorporate them into my script! I'm using r247 with my germany+iceland.poly file right now and it works as expected! Regards, Uli -- View this message in context: http://gis.19327.n5.nabble.com/splitter-r246-tp5737445p5737902.html Sent from the Mkgmap Development mailing list archive at Nabble.com.

Merge-line and routingprobleme was Not the best idealen the past. What about now ? Should we use it ? Am Montag, 26. November 2012 schrieb UliBaer :
Uli, can you please retry and give the exact error message of mkgmap?
I guess that one of your tile contains a large polygon that is now contained completely in the tile due to keep-complete. Maybe a part of mkgmap has a problem with this large polygon (e.g. subdivision handling). But this is wild guessing unless you can provide the exact error message. Wee need all information about your steps to create the germany+iceland.osm.pbf and your mkgmap parameters and style file so that we can try to reproduce the error.
Thanks WanMil
Hi WanMil, i described the way to produce the germany+iceland.osm.pbf in this post: http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2012q4/015462.html
I tried again using dataset from today with following parameters:
Splitter call: start /low /b /wait java -Xmx1348m -jar ..\bin\splitter.jar --max-threads=2 --keep-complete --overlap=0 --output=pbf --max-areas=256 --mapid=66%FID%001 --max-nodes=1200000 ..\OSM-Data\%osmfile%
mkgmap call: start /low /b /wait java -Xmx1348m -jar ..\bin\mkgmap.jar --max-jobs=3 --generate-sea=no-sea-sectors,extend-sea-sectors --country-name="%country%" --family-id=%FID% --mapname=66%FID%001 --overview-mapname=66%FID%000 --series-name="OSM-%country%" --family-name="OpenStreetMap: %country%" --style-file=..\bin\styles\ --style=%style% --description="%country%" -c ..\bin\template_basic.args -c template.args %TypFileToAdd%
template_basic.args: gmapsupp region-name:OSM latin1 product-id:1 area-name:OSM tdbfile route index bounds:..\bin\bounds.zip location-autofill:bounds,is_in,nearest coastlinefile=..\bin\coastlines_europe.osm.pbf delete-tags-file:..\bin\Tags_to_delete reduce-point-density-polygon:5 reduce-point-density:4 merge-lines min-size-polygon:8 remove-short-arcs:3 adjust-turn-headings link-pois-to-ways code-page:1252
Tags_to_delete: highway=proposed
The mkgmpap message is as follows: There is not enough room in a single garmin map for all the input data The .osm file should be split into smaller pieces first.
That's all, no further details and mkgmap continues with the other tiles. The defective tile with size 0 is 66923055.osm.pbf
Regards, Uli _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk <javascript:;> http://lists.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
-- Sent from Gmail Mobile
participants (6)
-
flabot@googlemail.com
-
Gerd Petermann
-
GerdP
-
Henning Scholland
-
UliBaer
-
WanMil