Tiles that crash MapSource with latest Mkgmap
I've been busy generating new Garmin tiles, but unfortunately some of the tiles cause MapSource to crash... Software used: - MapSource 6.13.7 - Mkgmap-r991 & r977 & r973 - Splitter.jar Commandline: java -Xmx2048M -jar ~/garmin/utils/mkgmap-r973/mkgmap.jar --latin1 --description='OSM world Routable' --net --route --series-name='OSM World Routable' 63240101.osm.gz I provided the max-nodes=1000000 (1 million) to the Splitter commandline to prevent too many outOfHeapSpace errors. This seemed to help because Mkgmap failed to generate a map for only three tiles for the whole of Europe, Asia, Africa and Oceania, all other tiles did not produce errors. Unfortunately, not having errors does not mean those tiles work flawlessly in MapSource: some cause it to crash instead while working fine on my GPSmap 60CSx. Some further testing showed that the tile was rendered (using the same commandline!) without causing the problems using Mkgmap-r830. A few of the problem tiles are: http://planetosm.oxilion.nl/~lambertus/r991/63240101.img http://planetosm.oxilion.nl/~lambertus/r991/63240103.img http://planetosm.oxilion.nl/~lambertus/r991/63240105.img All three are in dense southwest part of the Netherlands. The only workaround I can think about is to cut those tiles in half and try again... It would be nice to know the source of this problem and how to avoid it and also have Mkgmap to indicate clearly if there is a problem.
Some further testing showed that the tile was rendered (using the same commandline!) without causing the problems using Mkgmap-r830.
One way to find the error would be to narrow down the revision number by trying out e.g. r900. Afterwards try r860 or r950, depending on your results. And so on... May be a lot of work.... :-/
I would like to test, but I don't have those versions and don't know how I can create then from SVN. Johann Gail wrote:
Some further testing showed that the tile was rendered (using the same commandline!) without causing the problems using Mkgmap-r830.
One way to find the error would be to narrow down the revision number by trying out e.g. r900. Afterwards try r860 or r950, depending on your results. And so on...
May be a lot of work.... :-/
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
On Fri, 2009-04-03 at 23:02 +0200, Lambertus wrote:
I would like to test, but I don't have those versions and don't know how I can create then from SVN.
svn update -r <revision-number> e.g. $ svn up -r 972 U test/uk/me/parabola/mkgmap/general/PointInShapeTest.java ... U src/uk/me/parabola/mkgmap/general/MapRoad.java U src/uk/me/parabola/mkgmap/general/MapShape.java Updated to revision 972. Jon
It's not that difficult, if you have some base knowledge in programming. The process for building from the latest source is described at http://wiki.openstreetmap.org/wiki/Mkgmap/dev If you want to get some older sources, just do svn up -r950 ant dist to get the source for Revision 950 and compile it. Lambertus schrieb:
I would like to test, but I don't have those versions and don't know how I can create then from SVN.
Johann Gail wrote:
Some further testing showed that the tile was rendered (using the same commandline!) without causing the problems using Mkgmap-r830.
One way to find the error would be to narrow down the revision number by trying out e.g. r900. Afterwards try r860 or r950, depending on your results. And so on...
May be a lot of work.... :-/
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev .
Johann Gail wrote:
It's not that difficult, if you have some base knowledge in programming. The process for building from the latest source is described at http://wiki.openstreetmap.org/wiki/Mkgmap/dev
If you want to get some older sources, just do
svn up -r950 ant dist
to get the source for Revision 950 and compile it.
That wiki page is very useful indeed, thanks. I'll do some more testing.
On Thu, Apr 02, 2009 at 09:47:26PM +0200, Lambertus wrote:
I've been busy generating new Garmin tiles, but unfortunately some of the tiles cause MapSource to crash...
I've tried to make some similar maps and I think the reason is that the NET section is overflowing. If assertions are enabled (java -ea ... ) then I get this one triggered. assert offsetNet1 < 0x400000 : "NET 1 offset too large"; If is leave assertions not enabled then I get a map that appears to be broken in qlandkarte in a similar way to the maps you produced. So I guess we need to make this a proper error. And perhaps investigate why the NET section is sometimes so big.
Some further testing showed that the tile was rendered (using the same commandline!) without causing the problems using Mkgmap-r830.
That will be because routing with .osm files was not in r830 :) ..Steve
Sorry that I didn't respond earlier. I've been away for most of this weekend. Do you recommend to use the -ea option in my mapbuilding process just to notice that a map has failed? I certainly would like to know if it did, so I can flag it as such. Or does it do other (possibly unwanted) things as well? Thanks for looking into this. Do you still want me to try find out in which Mkgmap version this error occurs first? Steve Ratcliffe wrote:
On Thu, Apr 02, 2009 at 09:47:26PM +0200, Lambertus wrote:
I've been busy generating new Garmin tiles, but unfortunately some of the tiles cause MapSource to crash...
I've tried to make some similar maps and I think the reason is that the NET section is overflowing.
If assertions are enabled (java -ea ... ) then I get this one triggered.
assert offsetNet1 < 0x400000 : "NET 1 offset too large";
If is leave assertions not enabled then I get a map that appears to be broken in qlandkarte in a similar way to the maps you produced.
So I guess we need to make this a proper error. And perhaps investigate why the NET section is sometimes so big.
Some further testing showed that the tile was rendered (using the same commandline!) without causing the problems using Mkgmap-r830.
That will be because routing with .osm files was not in r830 :)
..Steve _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Hi
Do you recommend to use the -ea option in my mapbuilding process just to notice that a map has failed? I certainly would like to know if it did, so I can flag it as such. Or does it do other (possibly unwanted) things as well?
It would certainly be useful to use -ea. In theory assertions should only be triggered because of a bug or something unexpected or something that is not handled. In all those cases it would be good to know about the problem anyway.
Thanks for looking into this. Do you still want me to try find out in which Mkgmap version this error occurs first?
No I don't think that will be useful any more. It looks like the splitter is going to have to look at the number of ways or even the number of highways to deal with this without making all tiles smaller than they need to be. ..Steve
Steve Ratcliffe wrote:
Do you recommend to use the -ea option in my mapbuilding process just to notice that a map has failed? I certainly would like to know if it did, so I can flag it as such. Or does it do other (possibly unwanted) things as well?
It would certainly be useful to use -ea. In theory assertions should only be triggered because of a bug or something unexpected or something that is not handled. In all those cases it would be good to know about the problem anyway.
Thanks.
It looks like the splitter is going to have to look at the number of ways or even the number of highways to deal with this without making all tiles smaller than they need to be.
I'm looking forward to this.
participants (4)
-
Johann Gail -
Jon Burgess -
Lambertus -
Steve Ratcliffe