Re: [mkgmap-dev] Generating sea
Hi Wanmil, where can I get the executable builds (URL)? Greetings Josef Am 17.02.2011 18:24, schrieb WanMil: > Hi Josef, > > I can reproduce the sea problem. It has something to do with the > extend-sea-sectors algorithm. I don't know when I have time to find the bug. > > If you like you might help me to test the following mkgmap revisions: > <1682, 1684, 1704, 1712. Most important is 1712. 1682 and 1684 are from > the binfmt branch. > > You only need to compile the tile with the problem (it's 3007 in my > splitted tiles). > > Have fun! > WanMil > >> Hi Wanmil, >> >> No findings? >> >> The screenshots show clearly the difference between the maps created >> with mkgpmap-r1625 (generating sea is good, except the ghost rivers) and >> newer mkgmap-versions (incl. -r1846). >> >> Greetings >> Josef >> >> >> >> Am 10.02.2011 19:17, schrieb Josef Latt: >>> Hi Wanmil, >>> >>> Did you download the screenshots? >>> The missing north sea in sea_good.jpg is IMHO normal. It's outside the >>> bounding polygon. >>> >>> Thanks for your answer. >>> Josef >>> >>> >>> Parameters for splitter and mkgmap: >>> >>> java -ea -Xmx2560M -jar ./splitter/splitter.jar >>> --mapid=3001 --max-nodes=80 germany.osm >>> >>> java -ea -Xmx2560M -jar ./mkgmap-r1829/mkgmap.jar \ >>> --description=OSM_DE \ >>> --country-name=Deutschland --country-abbr=DE \ >>> --name-tag-list="name:de,name,int_name" \ >>> --style-file=./style/ \ >>> --generate-sea=extend-sea-sectors,close-gaps=5000 \ >>> --family-id= --product-id=1 \ >>> --series-name=OSM_Deutschland \ >>> --area-name=Deutschland \ >>> --overview-mapname=OSM_Deutschland \ >>> --tdbfile \ >>> --draw-priority=25 \ >>> --latin1 \ >>> --ignore-maxspeeds \ >>> --ignore-turn-restrictions \ >>> --no-poi-address \ >>> --max-jobs=2 \ >>> -c template.args .TYP >>> >>> >>> >>> Am 10.02.2011 18:45, schrieb WanMil: Josef, I downloaded the latest German extract from cloudmade.com. I splitted the extract using the default settings. In the map most of the North Sea is not created. Although this is not good I cannot reproduce your exact problem. Please provide the following information: * Splitter parameters * Your exact mkgmap parameters WanMil > Hi, > > I use the German extract from cloudmade.com. With mkgmap-r1815, 1827, > 1829 generating sea (--generate-sea=extend-sea-sectors,close-gaps=5000) > is incomplete. See screenshot sea_bad.jpg. > > Using mkgmap-r1625 generating sea is o.k. but there are some ghost > rivers. See screenshots sea_good.jpg and Ghostriver.jpg. > > Vou can download the screenshots as ZIP here: > http://ge.tt/#70BMn5z > > Greetings > Josef > ___ 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 > -- PGP Schlüssel: 311D1055 http://keyserver.pgp.com ___ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Re: [mkgmap-dev] Building the splitter with Eclipse?
On Thu, Feb 17, 2011 at 1:59 PM, Jeffrey Ollie wrote: > > I'm sure this is probably just a Java newbie thing... I only really > deal with Java for OSM so it's taking me a little while to get > accustomed to the tools. I'd like to use Eclipse as an editor because > of the inline help that it can give you, as opposed to Emacs which is > my usual editor. Grr, yes this is just a Java newbie thing... It seems that if you specify a -jar option on the command line, Java will ignore any -classpath option on the command line in favor of using a classpath specified inside the MANIFEST.MF. I switched to adding splitter.jar to the classpath and specifying the main class instead of using the -jar option. -- Jeff Ollie ___ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
[mkgmap-dev] NPE in splitter crosby_integration branch
I think that I've tracked down an NPE in the crosby_integration branch of the splitter. The SplitProcessor is using a BlockingQueue to queue up elements for writing. The poll() method can potentially return a null if the queue is empty. My question is, should the poll() method call be switched to a take() method call - the take() method will block if the queue is empty, or should I guard the usage of elements for null like in the patch below? Index: src/uk/me/parabola/splitter/SplitProcessor.java === --- src/uk/me/parabola/splitter/SplitProcessor.java (revision 161) +++ src/uk/me/parabola/splitter/SplitProcessor.java (working copy) @@ -320,13 +320,14 @@ } else { synchronized (workPackage) { while (!workPackage.inputQueue.isEmpty()) { - ArrayList elements =null; + ArrayList elements = null; try { elements = workPackage.inputQueue.poll(); - for (Element element : elements ) { - processElement(element, workPackage.writer); + if (elements != null) { + for (Element element : elements ) { + processElement(element, workPackage.writer); + } } - } catch (IOException e) { throw new RuntimeException("Thread " + Thread.currentThread().getName() + " failed to write element ", e); } -- Jeff Ollie ___ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
[mkgmap-dev] Building the splitter with Eclipse?
Can anyone give me tips on building the splitter (the crosby integration branch) with Eclipse? Whenever I try I end up with a jar file that can't find any of the dependencies on the classpath, even though I specify all of the external jars with the '-classpath' argument. I had similar problems building from the command line with ant until I modified the build.xml to copy all of the class files from the external jars into the splitter.jar file. I'm sure this is probably just a Java newbie thing... I only really deal with Java for OSM so it's taking me a little while to get accustomed to the tools. I'd like to use Eclipse as an editor because of the inline help that it can give you, as opposed to Emacs which is my usual editor. -- Jeff Ollie ___ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Re: [mkgmap-dev] --make-cycleways labels invisible
PS please ignore the tag "opposite-cycleway!=yes" that was just a test if rule one was applied don't use rule 2, because I thought maybe in rule 1) if oneway is set to -1 it automatically implies that the next rule will be executed too? It should be: highway=unclassified & oneway=yes & (cycleway=opposite | cycleway=opposite_lane) {set oneway=-1; set access=no; set bicycle=yes } [0x06 road_class=3 road_speed=5 resolution 22 continue ] highway=unclassified & oneway=-1 & (cycleway=opposite | cycleway=opposite_lane) {set oneway=yes; set access=no; set bicycle=yes} [0x06 road_class=3 road_speed=5 resolution 22 continue] ___ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Re: [mkgmap-dev] --make-cycleways labels invisible
I have tested this in the style file: highway=unclassified & oneway=yes & (cycleway=opposite | cycleway=opposite_lane) {set oneway=-1; set access=no; set bicycle=yes } [0x06 road_class=3 road_speed=5 resolution 22 continue ] highway=unclassified & oneway=-1 & opposite-cycleway!=yes & (cycleway=opposite | cycleway=opposite_lane) {set oneway=yes; set access=no; set bicycle=yes} [0x06 road_class=3 road_speed=5 resolution 22 continue] highway=unclassified [0x06 road_class=3 road_speed=5 resolution 22] This is not working. Routing is messed up terribly, especially it goes wrong where several oneway streets with "cycleway=opposite" cross each other. Often oneway streets are routable in the opposite direction, but not in the "oneway" direction anymore. I tried "continue with_actions" instead of "continue" too but it doesnt seem to help. I tried it without the oneway=-1 rule too but it still didn't work. It would be great if someone can provide me with a style file where this should work, I'm willing to test it out on this area: http://www.openstreetmap.org/?lat=52.157498&lon=5.394098&zoom=18&layers=M Regards, Minko ___ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Re: [mkgmap-dev] Generating sea
Hi Josef, I can reproduce the sea problem. It has something to do with the extend-sea-sectors algorithm. I don't know when I have time to find the bug. If you like you might help me to test the following mkgmap revisions: <1682, 1684, 1704, 1712. Most important is 1712. 1682 and 1684 are from the binfmt branch. You only need to compile the tile with the problem (it's 3007 in my splitted tiles). Have fun! WanMil > Hi Wanmil, > > No findings? > > The screenshots show clearly the difference between the maps created > with mkgpmap-r1625 (generating sea is good, except the ghost rivers) and > newer mkgmap-versions (incl. -r1846). > > Greetings > Josef > > > > Am 10.02.2011 19:17, schrieb Josef Latt: >> Hi Wanmil, >> >> Did you download the screenshots? >> The missing north sea in sea_good.jpg is IMHO normal. It's outside the >> bounding polygon. >> >> Thanks for your answer. >> Josef >> >> >> Parameters for splitter and mkgmap: >> >> java -ea -Xmx2560M -jar ./splitter/splitter.jar >> --mapid=3001 --max-nodes=80 germany.osm >> >> java -ea -Xmx2560M -jar ./mkgmap-r1829/mkgmap.jar \ >> --description=OSM_DE \ >> --country-name=Deutschland --country-abbr=DE \ >> --name-tag-list="name:de,name,int_name" \ >> --style-file=./style/ \ >> --generate-sea=extend-sea-sectors,close-gaps=5000 \ >> --family-id= --product-id=1 \ >> --series-name=OSM_Deutschland \ >> --area-name=Deutschland \ >> --overview-mapname=OSM_Deutschland \ >> --tdbfile \ >> --draw-priority=25 \ >> --latin1 \ >> --ignore-maxspeeds \ >> --ignore-turn-restrictions \ >> --no-poi-address \ >> --max-jobs=2 \ >> -c template.args .TYP >> >> >> >> Am 10.02.2011 18:45, schrieb WanMil: >>> Josef, >>> >>> I downloaded the latest German extract from cloudmade.com. I splitted >>> the extract using the default settings. In the map most of the North Sea >>> is not created. Although this is not good I cannot reproduce your exact >>> problem. >>> >>> Please provide the following information: >>> * Splitter parameters >>> * Your exact mkgmap parameters >>> >>> WanMil >>> Hi, I use the German extract from cloudmade.com. With mkgmap-r1815, 1827, 1829 generating sea (--generate-sea=extend-sea-sectors,close-gaps=5000) is incomplete. See screenshot sea_bad.jpg. Using mkgmap-r1625 generating sea is o.k. but there are some ghost rivers. See screenshots sea_good.jpg and Ghostriver.jpg. Vou can download the screenshots as ZIP here: http://ge.tt/#70BMn5z Greetings Josef >>> >>> ___ >>> 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
Re: [mkgmap-dev] --make-cycleways labels invisible
Thanks Henning, that is clear now. Most of those roads are found at residential and unclassified highways. Some tertiary and secondary, a few on primary roads and tracks with cycleway=opposite are rare. http://tagwatch.stoecker.eu/Netherlands/De/tagstats_highway_unclassified.html -- Henning wrote: You have to use a routable Garmin-way-ID instead of ?? and for ? you have to insert a numeric value (as normal). If you need several pairs of these rules depends on your generel style. If you want to distinguish eg. a tertiary oneway from a track oneway you need for each highway a pair of rules. ___ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Re: [mkgmap-dev] --make-cycleways labels invisible
Am 17.02.2011 00:01, schrieb Minko: > It's still a bit confusing to me. > Does ?? in [0x?? road_speed=? road_class=? continue] > mean that you can use wildcards? > > Or is it just an example and you need to fill it in for every road type, like > Felix already mentioned? > > And I suppose you need to add cycleway=opposite_lane as well. > And maybe bicycle:oneway=no or oneway:bicycle=no since some will use these > tags instead > of cycleway=opposite You have to use a routable Garmin-way-ID instead of ?? and for ? you have to insert a numeric value (as normal). If you need several pairs of these rules depends on your generel style. If you want to distinguish eg. a tertiary oneway from a track oneway you need for each highway a pair of rules. Henning ___ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev