Wrong_route_calculation / strange straight routing lines on GPS
On Jan 28, 2009, at 23:10, Gert Münzel wrote: /> making my first steps with mkgmap that, although the routing works />/> fine in Mapsource, i get in trouble on my Gps unit Garmin gmap60csx. />/> Very often i got these strange straight magenta routing lines in />/> between normal routing. />/> Meanwhile i figured out that it seems to be that t has something to />/> do with the length of a polyline. I don't know if it is the count of />/> points/nodes of the polyline or still the lenght or both. />/> But comparing the same mp-file once compiled with mkgmap, once with />/> cgpsmapper i see, that cgpsmapper split these long lines during />/> mkgmap leave them like they are. /> How did you compare the IMG files? Have you tried decompiling the broken IMG with test.display.*? Have you tried manually splitting the polylines in the .mp-file? Does it help?
My guess is that the problem lies with long polylines (in the .mp- file) that are split into several polylines (in the IMG) by mkgmap, which results in a road made up of several polylines.
But if I understand you correctly, maybe the problem is just with long polylines in the IMG file?
I think there must be something else than length of polylines and number of nodes. I get this error in a route that always starts at a node in a roundabout, no matter how far you fix the destination out of this point or in how many pieces you split the way. In another case, error seems to me to be related with the route crossing boundaries of an area tagged as area=yes, landuse=industrial, place_name=Pol. Ind. Las Capellanías
Carlos Dávila schrieb:
On Jan 28, 2009, at 23:10, Gert Münzel wrote: /> making my first steps with mkgmap that, although the routing works />/> fine in Mapsource, i get in trouble on my Gps unit Garmin gmap60csx. />/> Very often i got these strange straight magenta routing lines in />/> between normal routing. />/> Meanwhile i figured out that it seems to be that t has something to />/> do with the length of a polyline. I don't know if it is the count of />/> points/nodes of the polyline or still the lenght or both. />/> But comparing the same mp-file once compiled with mkgmap, once with />/> cgpsmapper i see, that cgpsmapper split these long lines during />/> mkgmap leave them like they are. /> How did you compare the IMG files? Have you tried decompiling the broken IMG with test.display.*? Have you tried manually splitting the polylines in the .mp-file? Does it help?
My guess is that the problem lies with long polylines (in the .mp- file) that are split into several polylines (in the IMG) by mkgmap, which results in a road made up of several polylines.
But if I understand you correctly, maybe the problem is just with long polylines in the IMG file?
I think there must be something else than length of polylines and number of nodes. I get this error in a route that always starts at a node in a roundabout, no matter how far you fix the destination out of this point or in how many pieces you split the way. In another case, error seems to me to be related with the route crossing boundaries of an area tagged as area=yes, landuse=industrial, place_name=Pol. Ind. Las Capellanías
Hi, Carlos, actually i don't us roundabouts i my trailmap. I have a full routable version of this map created with cgpsmapper. The database is the same mp-file. I guess not the count of pieces is important but there seems to be a max for length and/or cout of nodes. Also i guess it's more a coincidence that you observe the problem starting with a roundabout. I guess it's more like, if there are any polylines in any place of a calculated route, doesn't matter which type of polyline, which ar to long, you get in trouble with this route. Regarding the tagging of areas, i'm sorry but yet i don't know anything about osm-format. I think if the the data is in correctmp-formam the results with mkgmap and cgpsmapper should be the same, and as i already mention my map compiled with cgpsmapper definitly works fine. Also i use only a few small polygons for areas, which are not crossed by the route. Ok, anyway maybe there are more then one reason for this problem. cheers Gert
Gert Münzel escribió:
Carlos Dávila schrieb:
On Jan 28, 2009, at 23:10, Gert Münzel wrote: /> making my first steps with mkgmap that, although the routing works />/> fine in Mapsource, i get in trouble on my Gps unit Garmin gmap60csx. />/> Very often i got these strange straight magenta routing lines in />/> between normal routing. />/> Meanwhile i figured out that it seems to be that t has something to />/> do with the length of a polyline. I don't know if it is the count of />/> points/nodes of the polyline or still the lenght or both. />/> But comparing the same mp-file once compiled with mkgmap, once with />/> cgpsmapper i see, that cgpsmapper split these long lines during />/> mkgmap leave them like they are. /> How did you compare the IMG files? Have you tried decompiling the broken IMG with test.display.*? Have you tried manually splitting the polylines in the .mp-file? Does it help?
My guess is that the problem lies with long polylines (in the .mp- file) that are split into several polylines (in the IMG) by mkgmap, which results in a road made up of several polylines.
But if I understand you correctly, maybe the problem is just with long polylines in the IMG file?
I think there must be something else than length of polylines and number of nodes. I get this error in a route that always starts at a node in a roundabout, no matter how far you fix the destination out of this point or in how many pieces you split the way. Checking the map in gpsmapedit I realized that although I split the problematic road in several small pieces in JOSM, they are again merged in a single one when I create the mp file using osm2mp. I would like to split the road in the mp map to test again, but I don't find any tool for that in gpsmapedit. In another case, error seems to me to be related with the route crossing boundaries of an area tagged as area=yes, landuse=industrial, place_name=Pol. Ind. Las Capellanías
Hi, Carlos,
actually i don't us roundabouts i my trailmap. I have a full routable version of this map created with cgpsmapper. The database is the same mp-file. I guess not the count of pieces is important but there seems to be a max for length and/or cout of nodes. Also i guess it's more a coincidence that you observe the problem starting with a roundabout. I guess it's more like, if there are any polylines in any place of a calculated route, doesn't matter which type of polyline, which ar to long, you get in trouble with this route.
Regarding the tagging of areas, i'm sorry but yet i don't know anything about osm-format. I think if the the data is in correctmp-formam the results with mkgmap and cgpsmapper should be the same, and as i already mention my map compiled with cgpsmapper definitly works fine. Also i use only a few small polygons for areas, which are not crossed by the route. Ok, anyway maybe there are more then one reason for this problem.
cheers Gert
On Sat, Jan 31, 2009 at 10:05:26PM +0100, Carlos Dávila wrote:
problematic road in several small pieces in JOSM, they are again merged in a single one when I create the mp file using osm2mp. I would like to
osm2mp merges roads with the same name, so if you change the name of each piece then it will not merge them, I believe. But mkgmap already has code to split long lines as there is a limit of around 256 on the number of points in a line that the devices accept. Reducing the number MAX_POINTS_IN_LINE in LineSplitterFilter.java will reduce the line length. It could be that there is a limit to the number of routing nodes in the line not just the number of points. Regards ..Steve
Steve Ratcliffe escribió:
On Sat, Jan 31, 2009 at 10:05:26PM +0100, Carlos Dávila wrote:
problematic road in several small pieces in JOSM, they are again merged in a single one when I create the mp file using osm2mp. I would like to
osm2mp merges roads with the same name, so if you change the name of each piece then it will not merge them, I believe.
I tried changing the category of part of the road and now it's not merged. Original road had 335 nodes and 50.41 km. I split a portion of 83 nodes/13.26km where I had the problem in the route calculation and now this portion works fine, but the problem has moved to the remaining one. Note number of nodes is approximative, as gpsmapedit gives a different number depending on the zoom level.
But mkgmap already has code to split long lines as there is a limit of around 256 on the number of points in a line that the devices accept. Reducing the number MAX_POINTS_IN_LINE in LineSplitterFilter.java will reduce the line length.
It could be that there is a limit to the number of routing nodes in the line not just the number of points.
What do you mean by "routing nodes"? Are they nodes where to ways connect? Regards Carlos
On Feb 1, 2009, at 17:32, Carlos Dávila wrote:
What do you mean by "routing nodes"? Are they nodes where to ways connect?
Yes. They're noted in the .mp-file with the Nodi=x,y,z lines. The routing graph consists of these nodes and connections between them. Cheers Robert
participants (4)
-
Carlos Dávila -
Gert Münzel -
Robert Vollmert -
Steve Ratcliffe