March 2014 - mkgmap-dev - The mkgmap lists

link-pois-to-ways tags
by Gerd Petermann 22 Mar '14

22 Mar '14

Hi all, a few weeks ago we discussed the tags which should trigger the link-pois-to-ways option. The current implementation works like this: Each node with a tag access=*, barrier=*, or highway=* is handled as a POI (before style processing). Ways with such a node are tagged with mkgmap:way-has-pois=true. If the style assigns mkgmap:road-class=* or mkgmap:road-speed=* or one of the access tags (mkgmap:bicycle,mkgmap:foot,...,mkgmap:delivery) the node may change routing. WanMil suggested to remove access=* from the list of triggers, as it is too likely to be something like amenity=parking, access=private . This was forgotten during the merge of the mergeroads branch, so I've done it now with r3115. Gerd

5 19

mkgmap and route restrictions
by Gerd Petermann 21 Mar '14

21 Mar '14

Hi all, looking at the 4 node restrictions I think I found a general bug in mkgmap: Both with polish (.mp) format and with OSM data we compute the routing nodes for which the restriction shoud be applied. In a second step we compute the arcs between these nodes. In this 2nd step we ignore the information about roads given with TRAFFROADS in polish format or the from,via, and to roles in OSM data. In the mkgmap algo, we just select the first direct arc (means: peace of road) between the nodes. If the nodes are connected with multiple arcs on different roads, this probably is too lazy, although it will not fail very often. I have to rewrite most of the code reg. restrictions to correct this.. Gerd

5 8

weird behavior of contour lines
by Michał Rogala 19 Mar '14

19 Mar '14

Hi! I generated my map, merged with SRTM (using splitter) and my complilation has a lot of blank rectangles in contour lines: http://imgur.com/7zP0TNt Did anyone see something like that? Is it phyghtmap problem or something with mkgmap/splitter? Best regards Michal Rogala

2 6

Feedback on mkgmap r3111
by Thorsten Kukuk 19 Mar '14

19 Mar '14

Hi, during the last two days I had the chance to compare a lot of long distance routing of my Garmin 62s (Firmware 5.30, so newest one with activity routing) with my map generated with mkgmap r3111 and a commercial built in car navigation system. The routing on long distance between both devices was the same. The difference between the km the 62s calculated and we drove was about 1%, the time about 3%. Inside of cities on short routes the 62s did choose the better routes, but I think that's only because the OSM data was more up-to-date. So the "NOD127" branch was really a very good job! There were only two negative aspects: - the 5.30 firmware for the 62s is a nightmare in stability and usability :( - if you enable "avoid u-turns", the 62s will not only avoid them, but really never does them, even if it means there is no legal way to continue. Means in two situations we were suggested to use a cycleway or a footway. That's reproduceable with the 62s, but not with MapSource, so I think it's a firmware bug. (if somebody wants to try: Route to "Tigerpalast" in Frankfurt a. Main and continue with Routing to "Lange Strasse 30-36" with "avoid u-turn" enabled, car, fastest route). Thorsten -- Thorsten Kukuk, Senior Architect SLES & Common Code Base SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)

2 1

Bogus roundabout warnings
by Marko Mäkelä 18 Mar '14

18 Mar '14

With the most recent mkgmap trunk (r3111) I am getting bogus warnings like these: 2014/03/14 22:07:19 WARNING (RouteNode): 63240008.osm.pbf: Roundabout (http://www.openstreetmap.org/browse/way/32502891) forks at http://www.openstreetmap.org/?mlat=63.830711&mlon=23.120524&zoom=17 2014/03/14 22:07:19 WARNING (RouteNode): 63240008.osm.pbf: Outgoing roundabout flare road (http://www.openstreetmap.org/browse/way/27506601) does not finish at flare? http://www.openstreetmap.org/?mlat=63.905483&mlon=24.516046&zoom=17 2014/03/14 22:07:19 WARNING (RouteNode): 63240008.osm.pbf: Roundabout (http://www.openstreetmap.org/browse/way/27506649) forks at http://www.openstreetmap.org/?mlat=63.905981&mlon=24.514913&zoom=17 I suspect that these are being triggered when there are nodes very close to each other, and they are being merged. I do not remember seeing these warnings yesterday, with r3110. Marko

2 5

Commit: r3116: reduce severity level of "very small shape near ... " message
by svn commit 18 Mar '14

18 Mar '14

Version mkgmap-r3116 was committed by gerd on Tue, 18 Mar 2014 reduce severity level of "very small shape near ... " message from warning to debug.

1 0

Commit: r3115: don't use trigger tag access=* for link-pois-to-ways option
by svn commit 18 Mar '14

18 Mar '14

Version mkgmap-r3115 was committed by gerd on Tue, 18 Mar 2014 don't use trigger tag access=* for link-pois-to-ways option

1 0

HTTP Error 500: INTERNAL SERVER ERROR
by Bernd Weigelt 17 Mar '14

17 Mar '14

Hi Got this, if try to download a file from http://www.mkgmap.org.uk/download/mkgmap.html thx Bernd

2 1

is tag matching case sensitive?
by Michał Rogala 17 Mar '14

17 Mar '14

Hi! I've always thought that matching tag values is case insensitive. But today I found a situation where colour='#00ff00' rule doesn't match when tag value is '#00FF00'. creating a rule (colour='#00ff00' | colour='#00FF00') solved my problem. Is it a bug in mkgmap or this kind of matching should by design be case sensitive? Maybe the hash sign causes trouble somewhere. best regards Michal Rogala

3 4

Commit: r3114: fix roundabout checks: only use direct arcs
by svn commit 17 Mar '14

17 Mar '14

Version mkgmap-r3114 was committed by gerd on Mon, 17 Mar 2014 fix roundabout checks: only use direct arcs

1 0