[mkgmap-dev] highway=track with no road_class and road_speed set breaks routing ## was ## "No roads near target" bug in Schwabmünchen

Mon Apr 15 21:31:02 BST 2013

GerdP wrote
> I've used Bernds style to reproduce the problem:
> https://github.com/berndw1960/aiostyles
> 
> Attached is a patch that replaces the "mkgmap:check_connected"  feature
> with 
> a "mkgmap:set_unconnected_type" feature.
> routable_types_v3.patch
> <http://gis.19327.n5.nabble.com/file/n5757151/routable_types_v3.patch>  
> 
> Old version was:
> highway=service | highway=estate | highway=living_street
> 		{
> 		  add mkgmap:dead-end-check=false;
> 		  add mkgmap:check_connected = true
> 		}
> 
> the new rule should look similar to this:
> 
> highway=service | highway=estate | highway=living_street
> 		{
> 		  add mkgmap:dead-end-check=false;
> 		  add mkgmap:set_unconnected_type = 0x12007
> 		}
> 
> The meaning is: If the highway is not connected to any road , change the
> type
> to 0x12007 and add it as a line instead of a road.

Now I am a little lost.
In the lines file, just below the lines you mention above, there are already
rules that draw the 
highway=service,  highway=estate and highway=living_street with
non-routable-types. 
It's some kind of overlay if I understand that right (continue with
actions).
Then there is a line that adds the routable-type 0x07, which is made
invisible in the TYP file.
I suspect this line is only there for routing.

Now how does mkgmap decide which of rules it has to change or even drop if
"not connected" ?

Wouldn't it be much more straightforward to make "check_connected" add a
variable, let's say "is_connected" that could be used later on in the rules?

For a beginner like me the rules would be far more easy to read and
understand.

> Besides that the patch implements the changes of routable_types_v2.patch.
> The compiled binary is here:
> http://files.mkgmap.org.uk/download/107/mkgmap.jar
> 
> Gerd

I tested your binary with the proposed change on the Style, and now I still
get precisely 1 Error Message in my test case. Before there were plenty
(maybe 100) of them.

Ciao, 
  Franco

--
View this message in context: http://gis.19327.n5.nabble.com/No-roads-near-target-bug-in-Schwabmunchen-tp5753364p5757237.html
Sent from the Mkgmap Development mailing list archive at Nabble.com.