[mkgmap-dev] bugreport for new splitter
From Chris Miller chris.miller at kbcfp.com on Sun Aug 9 13:14:14 BST 2009
Hi Paul, Thanks for the detailed bug report, and sorry for the inconvenience this one has caused you. It looks like there's a bug in the new code for handling ways that span > 4 areas. I'll take a look shortly and hopefully have a fix later today. As an aside, I see you have set --max-areas=100. Given that you have a 4GB heap and --max-nodes at just 300000, you should easily be able to get away with the default of --max-areas=255. From my (limited) experiements so far, with a 4GB heap you shouldn't need to worry about reducing max-areas until your max-nodes gets up somewhere near 1,000,000. Setting it to a lower value will just cause the split to run slower. Chris > Hello! > > I have just testet the latest svn version of splitter (r64) I get the > following exception: > > 12,000,000 ways processed... > [GC 926169K->330369K(1422720K), 0.0038790 secs] > Writing relations Sun Aug 09 11:11:14 CEST 2009 > [GC 837769K->592513K(1423296K), 0.1517410 secs] > Exception in thread "main" java.lang.IndexOutOfBoundsException: > bitIndex < 0: -2147483627 > at java.util.BitSet.set(BitSet.java:262) > at > uk.me.parabola.splitter.SplitParser.processRelation(SplitParser.java:1 > 89) > at > uk.me.parabola.splitter.SplitParser.startElement(SplitParser.java:96) > at > uk.me.parabola.splitter.AbstractXppParser.parse(AbstractXppParser.java > :38) > at uk.me.parabola.splitter.Main.writeAreas(Main.java:231) > at uk.me.parabola.splitter.Main.split(Main.java:98) > at uk.me.parabola.splitter.Main.main(Main.java:78) > The splittter was called as follows: > java -Xmx3800m -verbose:gc -jar ../splitter.jar --max-areas=100 > --max-nodes=300000 ../../europe.osm > With todays (uncompressed) Europe extract from geofabrik. > > I have tried to use > java -Xmx3800m -verbose:gc -jar ../splitter.jar --max-areas=100 > --max-nodes=300000 ../../europe.osm > but after "Writing relations" the CPU was at 100% for an hour without > reporting anything on the console nor writing any data to disk. > I was waiting for a few months for updating my maps on the etrex... I > thought already, now it will be possible again... > Thanks a lot for fixing splitter. It seems to be really close! > I'll try to report any problems I encounter with new versions ASAP. > > Paul >
- Previous message: [mkgmap-dev] bugreport for new splitter
- Next message: [mkgmap-dev] bugreport for new splitter
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the mkgmap-dev mailing list