logo separator

[mkgmap-dev] mkgmap ToDo list

From Gerd Petermann gpetermann_muenchen at hotmail.com on Thu Apr 17 07:06:15 BST 2014

Hi Felix,

the old behaviour (before r2944 which was committed 2014-01-05) was that only the routable ways were reversed. 
The patch should bring back this behaviour. 

Gerd

Date: Thu, 17 Apr 2014 13:54:39 +0800
From: extremecarver at gmail.com
To: mkgmap-dev at lists.mkgmap.org.uk
Subject: Re: [mkgmap-dev] mkgmap ToDo list


  
    
  
  
    
      I currently have basically no time - so I'm just reading on
        the list in skip through mode. Unable to test - but I'm pretty
        sure there is a big bug related to the oneway handling and
        reversing of ways. I'm also quite sure it wasn't there last year
        and got introduced with the change of the style-file getting all
        tags instead of internal handling.

        

      
      My style is a big mess right now (trying to adapt to the changes
      needed, but fucking up because no set mkgmap:access=yes/no and
      delete mkgmap:access shortcut exists anymore amongst other
      imcompatibilities compared to before) -  but I'm pretty sure that
      the oneway handling - also related to oneway=yes or oneway=-1 set
      in the lines file somehow doesn't work correctly.

      

      I often reverse ways - in order to have depending on direction
      different classification of the way (regarding both layout and
      routing priority) and have noticed that lately I'm routed against
      the intended direction because mkgmap puts the way in the wrong
      direction. And I'm 95% sure that the error is on mkgmap not the
      style-file but didn't find the time to analyze...

      

    
    Being in China right now behind the great firewall - it's also
    cumbersome to access my SVN and other things as they are blocked - I
    won't be able to help much/at all before mid June in finding or
    specifying bugs...

    

    

    Alternative 1 sounds best - as it would be in line with the old
    behaviour if I'm not incorrectly understanding it here... 

    Or Alternative 3 - don't do anything!! Leave it to the style
    developer - wasn't this the old behaviour? 

    I don't know since when this is bugged - but I'm sure 1 year ago
    everything worked correctly if you assigned all ways to either
    oneway=yes or oneway=-1 (or was it reverse?). Only thing to
    normalize was true or 1 to yes and so on...

    

    

    On 17.04.2014 13:42, Gerd Petermann
      wrote:

    
    
      
      Hi Henning,

        

        okay. I think WanMil told me that before, but I forgot again ...
        

        

        A few details:

        

        mkgmap always reversed routable ways with oneway=-1 or
        oneway=reverse after style processing.

        

        With r2944 I've changed the code so that this happens before
        calling RoadMerger and also that overlaying 

        lines are reversed.

        

        If I got that right, the style may use different types for
        overlaying lines, depending on the oneway tag,

        so if you use the same style for r2943 or older and for current
        version, you should see a difference

        for oneway=-1 ways.

        I don't know how difficult it is to change the style to produce
        correct results with current version again,

        but you have to remember that the way is reversed, so left will
        be right, forward will be backward.

        

        I see two alternatives:

        1) mkgmap reverses the way before style processing. This
        requires also to reverse all tags like

        bicycle:left, car:forward etc, so it is not trivial. On the
        other hand, it might allow to merge more 

        roads in RoadMerger.

        2) mkgmap makes sure that overlaying lines are NOT reversed. I
        think this can be done

        with the attached patch.

        

        By the way:  There are also tags on nodes which depend on the
        direction of the way, see

http://gis.19327.n5.nabble.com/direction-forward-backward-on-nodes-tt5803072.html

        

        I think we can ignore this because in the points file you can't
        detect the direction of the way.

        

        Gerd

        
          Date: Wed, 16 Apr 2014 23:00:35 +0200

          From: osm at aighes.de

          To: mkgmap-dev at lists.mkgmap.org.uk

          Subject: Re: [mkgmap-dev] mkgmap ToDo list

          

          Am 16.04.2014 22:01, schrieb
            GerdP:

          
          
            Why do you think that it is a bug?
          
          Think

              about oneway=-1 and in style you add oneway-arrows (link
              mapnik does). In style you will handle -1 special, because
              arrows should drawn in oposite direction of way. If mkgmap
              now changes direction of the way, arrows should be in
              wrong direction, or am I wrong?

              

              Henning

             

          _______________________________________________
          mkgmap-dev mailing list
          mkgmap-dev at lists.mkgmap.org.uk
          http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
      
      

      
      

      _______________________________________________
mkgmap-dev mailing list
mkgmap-dev at lists.mkgmap.org.uk
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
    
    

  


_______________________________________________
mkgmap-dev mailing list
mkgmap-dev at lists.mkgmap.org.uk
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20140417/921c5931/attachment.html>


More information about the mkgmap-dev mailing list