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

Tue Apr 9 21:46:38 BST 2013

Hi Franco,

that is bad news. Maybe I did something wrong during my tests, or I didn't
post the correct binary.
I will double check the results tomorrow, please do the same.

Gerd

franco_bez wrote
> Hi Gerd,
> 
> I assume that I can skip the tests with the two versions 2447 and 2459 ?
>  
> I tested with the patched binary you provided.
> with the "bogus" style there is no big difference 

> On the Oregon I get "no roads near target" both for the POI and for the
> address, just like before.
> On the Nüvi I get "calculating xx%" forever also for both targets. At
> least it does no longer propose an impossible route over non existant
> ways.
> 
> In the mean time Bernd eliminated all routeable types without "road_speed"
> and "road_class" in the "real" style.
> Also in this style there is no noticeable change using your patched
> Version.
> Both on the Nüvi and the Oregon routing to the POI is OK in the "real"
> style, both with and without your patched version.
> Routing to the address is still broken, both the Nüvi and the Oregon show
> an impossible route over a non existant way, both with and without your
> patched version.
<http://gis.19327.n5.nabble.com/file/n5756442/5_OREGON_ROUTE.png> 
> 
> I'm not sure if your patched binary contains the warnings you implemented
> for routeable lines without road_speed ?
> this is the output of --list-styles using your patched version on the
> "real" stlye
>> java -jar mkgmap.jar 
>> --style-file=/home/franco/map_build/mystyles/basemap_style/ --list-styles
>> Time started: Tue Apr 09 21:29:47 CEST 2013
>> The following styles are available:
>> basemap_style      1.0: The default AIO-basemap-style
>> Time finished: Tue Apr 09 21:29:47 CEST 2013
>> Total time taken: 472ms
> 
> I assume that the routing Problem to the POI can be avoided by using a
> style where all routable types have the road_speed and road_class
> attributes.
> But routing to the address must have a second "trigger" in the style.
> 
> I will now start the search for the offending lines in the "real" style.
> 
> Ciao,
>   Franco
> GerdP wrote
>> Hi all,
>> 
>> attached is routable_types_v1.patch which (hopefully) corrects the last
>> routing problems introduced with r2448.
>> A compiled binary of trunk version r2562 (svnversion says 2564m) with
>> this patch is here:
>> http://files.mkgmap.org.uk/download/102/mkgmap.jar
>> 
>> Please try it f your style assigns routable types (0x01-0x13, 0x1a,0x1b)
>> to lines without
>> setting road_class or road_speed, else you should see no difference.
>> 
>> Gerd
>> GerdP wrote
>>> Hi,
>>> 
>>> thanks for the package and the hints regarding the test. I finally was
>>> able to reproduce the problem and found that the
>>> last version which worked with the "bogus" style file was probably
>>> r2447.
>>> r2248 contained my "removeShortArcs" patch, so it seems that there might
>>> be a need to remove short arcs not only from
>>> roads but maybe from all lines with routable types. I'll try to
>>> implement that now.
>>> 
>>> Gerd

--
View this message in context: http://gis.19327.n5.nabble.com/No-roads-near-target-bug-in-Schwabmunchen-tp5753364p5756443.html
Sent from the Mkgmap Development mailing list archive at Nabble.com.