Commit: r1377: Limit length of lines to work around problems when subdividing long lines.
Version 1377 was commited by markb on 2009-11-11 21:25:03 +0000 (Wed, 11 Nov 2009) Limit length of lines to work around problems when subdividing long lines. This is not the ideal solution but works as an interim fix until the real problem is solved. The fundamental problem appears to be that very long lines (national boundaries, etc.) can be segmented such that some of the line's segments fall outside the map area that the line has been assigned to. By chopping the line into segments before the line is assigned to a subdivision, the problem is avoided. This can't be done for polygons and so very large polygons can still yield errors.
svn commit wrote:
Version 1377 was commited by markb on 2009-11-11 21:25:03 +0000 (Wed, 11 Nov 2009)
Limit length of lines to work around problems when subdividing long lines.
This is not the ideal solution but works as an interim fix until the real problem is solved.
The fundamental problem appears to be that very long lines (national boundaries, etc.) can be segmented such that some of the line's segments fall outside the map area that the line has been assigned to. By chopping the line into segments before the line is assigned to a subdivision, the problem is avoided.
This can't be done for polygons and so very large polygons can still yield errors. _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
I am not sure which of todays changes is responsible, but routing in Mapsource became a lot worse! Often I get routes 10-30% longer or on less priorised ways than before, this is a drastic example but happens on every 10 route: Yellow route is old SVN 1773, pink is newest SVN version (yellow is 17.8km and 33 minutes, pink is 26.5km and 56 minutes):
Hi Felix,
I am not sure which of todays changes is responsible, but routing in Mapsource became a lot worse! Often I get routes 10-30% longer or on less priorised ways than before, this is a drastic example but happens on every 10 route:
Yellow route is old SVN 1773, pink is newest SVN version (yellow is 17.8km and 33 minutes, pink is 26.5km and 56 minutes):
Please try increasing the values of MAX_ARC_LENGTH and MAX_LINE_LENGTH and see if that improves things. I would not expect MAX_LINE_LENGTH to make a difference as that only influences lines not routable ways. If the reduction in MAX_ARC_LENGTH is causing your routing to become worse, that's going to conflict with the issue Marko reported. Cheers, Mark
Hi Mark, Felix, On Thu, Nov 12, 2009 at 09:30:17AM +0000, Mark Burton wrote:
Hi Felix,
I am not sure which of todays changes is responsible, but routing in Mapsource became a lot worse! Often I get routes 10-30% longer or on less priorised ways than before, this is a drastic example but happens on every 10 route:
Yellow route is old SVN 1773, pink is newest SVN version (yellow is 17.8km and 33 minutes, pink is 26.5km and 56 minutes):
Please try increasing the values of MAX_ARC_LENGTH and MAX_LINE_LENGTH and see if that improves things. I would not expect MAX_LINE_LENGTH to make a difference as that only influences lines not routable ways.
If the reduction in MAX_ARC_LENGTH is causing your routing to become worse, that's going to conflict with the issue Marko reported.
The Nuvi user reported that he is still having routing trouble with the MAX_ARC_LENGTH=39062 that I used in conjunction with r1366. I asked him to experiment with different values of MAX_ARC_LENGTH. You reduced the parameter to 35000, IIRC. Marko
Mark Burton wrote:
Hi Felix,
I am not sure which of todays changes is responsible, but routing in Mapsource became a lot worse! Often I get routes 10-30% longer or on less priorised ways than before, this is a drastic example but happens on every 10 route:
Yellow route is old SVN 1773, pink is newest SVN version (yellow is 17.8km and 33 minutes, pink is 26.5km and 56 minutes):
Please try increasing the values of MAX_ARC_LENGTH and MAX_LINE_LENGTH and see if that improves things. I would not expect MAX_LINE_LENGTH to make a difference as that only influences lines not routable ways.
If the reduction in MAX_ARC_LENGTH is causing your routing to become worse, that's going to conflict with the issue Marko reported.
Could it be that at resolution 24 we have filters running for smoothing lines that cause the problem, or the tweeze arc causing problems with low values? The crazy thing is that if I increase max_arc_length at resolution=24 I will even see roads with different shapes/angles at the junction. The thing is that with more max_arc_length and max_line_length (both set to 80.000) even on very short roads routing get better - roads that definitely don't have more than 20 points or longer than 200-300m. This however happens only once the overall length of the route without via points gets longer (starting at about 30-40km). Another thing we could try is to make the tweeze arc filter much more radical - maybe this would help in the routing problem. I still cannot believe that the max_arc_length itself is responsible for the problems on the Nuvi.
Cheers,
Mark _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Hi Felix,
I still cannot believe that the max_arc_length itself is responsible for the problems on the Nuvi.
No, I don't believe that either. There's something else going wrong. As I mentioned in my email to Marko, I have another patch that, essentially, all it does is remove nodes that don't need to be nodes and if I use that, it breaks routing on the Nuvi (but not on mapsource or the etrex). I wonder if that is related to the problem that Marko reports. Cheers, Mark
participants (4)
-
Felix Hartmann -
Mark Burton -
Marko Mäkelä -
svn commit