logo separator

[mkgmap-dev] Splitter problem: losing a track segment

From Alexandre Loss alexandre.loss at gmail.com on Tue Jun 2 13:45:10 BST 2015

Ok Gerd,

I'll include this change in the roadmap of my generator. You know as a
developer... there are some many things to do in the backlog and our time
available is so short! :-)
Thanks.

Alexandre

2015-06-02 9:34 GMT-03:00 GerdP <gpetermann_muenchen at hotmail.com>:

> Hi Alexandre,
>
> well, the advantage of the keep-complete option is that you don't need
> the large overlap and still get complete ways and relations, so
> it's about disk usage, memory and quality. As long as you files are rather
> small
> you can probably use a large overlap value instead.
> Maybe you can calculate the needed overlap in your generator.
>
> ciao,
> Gerd
>
>
> Alexandre Loss wrote
> > Hi Gerd,
> >
> > I already made a previous analysis before to change my generator. Well,
> > it's possible, but I practically have to rebuild the generator entirely.
> > So, I understood that you said that it's better solution but not a "must"
> > solution to change generator. If so, I will prefer to keep the generator
> > this way for a while.
> > However, If you think I can be missing something in my maps due this
> mixed
> > syntax, then I will plan to change my code.
> >
> > Thanks
> >
> > Alexandre
> >
> > 2015-06-02 2:36 GMT-03:00 Gerd Petermann <
>
> > gpetermann_muenchen@
>
> > >:
> >
> >> Hi Alexandre,
> >>
> >> the better solution would be to change your generator so that
> >> it writes the data in the expected order (nodes, ways, relations, each
> >> with
> >> ascending ids). If you do that, you can remove the options --mixed and
> >> --overlap
> >> and use the default --keep-complete=true.
> >>
> >> A possible compromise:
> >> I have to double check that, but I think the keep-complete option only
> >> requires that
> >> the ids are sorted, it doesn't care about the order of the elements.
> >> On the other hand, the xml parser requires the
> >> mixed option when the element types are not sorted,
> >> and a check prohibtis the combination of --mixed  and
> keep-complete=true.
> >> If I am right, I can remove this check.
> >>
> >> Ciao,
> >> Gerd
> >>
> >> ------------------------------
> >> Date: Sat, 30 May 2015 09:15:19 -0300
> >> From:
>
> > alexandre.loss@
>
> >> To:
>
> > mkgmap-dev at .org
>
> >> Subject: Re: [mkgmap-dev] Splitter problem: losing a track segment
> >>
> >>
> >> Hi Gerd,
> >>
> >> Overlap of 10000 fixed my problem. :-)
> >>
> >> Thank you very much!
> >>
> >> Alexandre
> >>
> >> 2015-05-30 8:20 GMT-03:00 Alexandre Loss <
>
> > alexandre.loss@
>
> > >:
> >>
> >> Hi Gerd,
> >>
> >> Thanks for your answer.
> >> I'll try your suggestion and give you a feedback about the result.
> >>
> >> Regards,
> >>
> >> Alexandre
> >>
> >> 2015-05-30 5:43 GMT-03:00 Gerd Petermann <
>
> > gpetermann_muenchen@
>
> > >
> >> :
> >>
> >> Hi Alexandre,
> >>
> >> I don't have much time to analyse your data today.
> >> I think splitter assumes that the tile boundaries
> >> are
> >> "multiples of 0x200 map units wide and high"
> >> when you use resolution=15.
> >>
> >> Since you use keep-complete=false you have to watch
> >> the overlap value.
> >> You can use a larger overlap value like 10000 to fix this problem.
> >>
> >> Gerd
> >>
> >>
> >> ------------------------------
> >> Date: Fri, 29 May 2015 22:04:34 -0300
> >> From:
>
> > alexandre.loss@
>
> >> To:
>
> > mkgmap-dev at .org
>
> >> Subject: [mkgmap-dev] Splitter problem: losing a track segment
> >>
> >>
> >> Hi guys,
> >>
> >> I'm facing a problem in one of my maps, that is probably might be
> >> happening in other places where I didn't see yet.
> >> The problem can be seen in the figure below taken from MapSource. There
> >> is
> >> a gap in a track on the border of the tiles:
> >>
> >> [image: Imagem inline 2]
> >>
> >> So I started to analyze the problem step-by-step till find that the
> >> problem was generated by Splitter (last version).
> >>
> >> To simplify your analysis, I edited my .osm file before running Splitter
> >> and remove all objects (nodes, lines, poi, etc.) not related to the
> >> problem, leaving only the line where the problem happens. The result was
> >> the file "93352150-Ribeirao_Preto.osm" attached, containing only this:
> >>
> >> <?xml version="1.0" encoding="UTF-8"?>
> >>
> > <osm version="0.6" generator="TSuite 6.6.0.4">
> >>
>
> >>
> > <node id="576" lat="-21.9388623" lon="-48.3347007" visible="true"
> >>
> >  version="1"/>
> >>
> > <node id="805" lat="-21.8877920" lon="-48.2932640" visible="true"
> >>
> >  version="1"/>
> >>
> > <node id="806" lat="-21.8674906" lon="-48.2767485" visible="true"
> >>
> >  version="1"/>
> >>
> > <node id="804" lat="-21.8661147" lon="-48.2756044" visible="true"
> >>
> >  version="1"/>
> >>
> > <way id="803" visible="true" version="1">
> >>
> > <nd ref="576"/>
> >>
> > <nd ref="805"/>
> >>
> > <nd ref="806"/>
> >>
> > <nd ref="804"/>
> >>
> > <tag k="highway" v="trunk"/>
> >>
> > <tag k="name" v="SP-255 DANIELLE"/>
> >>
> > <tag k="ref" v="SP-255"/>
> >>
> > <tag k="addr:city" v="BOA ESPERANÇA DO SUL"/>
> >>
> > <tag k="addr:state" v="SP"/>
> >>
> > <tag k="addr:country" v="Brasil"/>
> >>
> > </way>
> >>
>
> >>
> > </osm>
> >>
> >> Then I ran Splitter on this file and the result was the file
> >> "93352150.osm" (attached and compacted), whose content are the lines
> >> bellow:
> >>
> >> <?xml version='1.0' encoding='UTF-8'?>
> >>
> > <osm version='0.5' generator='splitter' upload='false'>
> >>
> > <bounds minlat='-21.895751953125' minlon='-50.262451171875'
> >>
> >  maxlat='-20.401611328125' maxlon='-47.548828125'/>
> >>
> > <node id='805' lat='-21.8877920' lon='-48.2932640'/>
> >>
> > <node id='806' lat='-21.8674906' lon='-48.2767485'/>
> >>
> > <node id='804' lat='-21.8661147' lon='-48.2756044'/>
> >>
> > <way id='803'>
> >>
> > <nd ref='576'/>
> >>
> > <nd ref='805'/>
> >>
> > <nd ref='806'/>
> >>
> > <nd ref='804'/>
> >>
> > <tag k='highway' v='trunk'/>
> >>
> > <tag k='name' v='SP-255 DANIELLE'/>
> >>
> > <tag k='ref' v='SP-255'/>
> >>
> > <tag k='addr:city' v='BOA ESPERANÇA DO SUL'/>
> >>
> > <tag k='addr:state' v='SP'/>
> >>
> > <tag k='addr:country' v='Brasil'/>
> >>
> > </way>
> >>
> > </osm>
> >>
> >> Note that in the "93352150.osm" file is missing the node 576 definition
> >> (i.e.:
> > <node id="576" lat="-21.9388623" lon="-48.3347007" visible="true"
> >>
> >  version="1"/>).
> >> And this is the cause of the lost segment.
> >>
> >> I'm running Splitter in the following way:
> >>
> >> java -jar splitter.jar --split-file=93352150-Ribeirao_Preto.list
> >> --max-areas=1024 --mixed=true --keep-complete=false --resolution=15
> >> --output=xml 93352150-Ribeirao_Preto.osm
> >>
> >> The above .list file is attached.
> >>
> >> If this is not a Splitter error, please, let me know what I'm doing
> >> wrong.
> >>
> >> Thanks.
> >>
> >> Alexandre
> >>
> >>
> >> _______________________________________________ mkgmap-dev mailing list
> >>
>
> > mkgmap-dev at .org
>
> >> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >>
> >> _______________________________________________
> >> mkgmap-dev mailing list
> >>
>
> > mkgmap-dev at .org
>
> >> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >>
> >>
> >>
> >>
> >> _______________________________________________ mkgmap-dev mailing list
> >>
>
> > mkgmap-dev at .org
>
> >> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >>
> >> _______________________________________________
> >> mkgmap-dev mailing list
> >>
>
> > mkgmap-dev at .org
>
> >> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >>
> >
> > _______________________________________________
> > mkgmap-dev mailing list
>
> > mkgmap-dev at .org
>
> > http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >
> > image.png (12K)
> > <http://gis.19327.n5.nabble.com/attachment/5846805/0/image.png>
>
>
> Alexandre Loss wrote
> > Hi Gerd,
> >
> > I already made a previous analysis before to change my generator. Well,
> > it's possible, but I practically have to rebuild the generator entirely.
> > So, I understood that you said that it's better solution but not a "must"
> > solution to change generator. If so, I will prefer to keep the generator
> > this way for a while.
> > However, If you think I can be missing something in my maps due this
> mixed
> > syntax, then I will plan to change my code.
> >
> > Thanks
> >
> > Alexandre
> >
> > 2015-06-02 2:36 GMT-03:00 Gerd Petermann <
>
> > gpetermann_muenchen@
>
> > >:
> >
> >> Hi Alexandre,
> >>
> >> the better solution would be to change your generator so that
> >> it writes the data in the expected order (nodes, ways, relations, each
> >> with
> >> ascending ids). If you do that, you can remove the options --mixed and
> >> --overlap
> >> and use the default --keep-complete=true.
> >>
> >> A possible compromise:
> >> I have to double check that, but I think the keep-complete option only
> >> requires that
> >> the ids are sorted, it doesn't care about the order of the elements.
> >> On the other hand, the xml parser requires the
> >> mixed option when the element types are not sorted,
> >> and a check prohibtis the combination of --mixed  and
> keep-complete=true.
> >> If I am right, I can remove this check.
> >>
> >> Ciao,
> >> Gerd
> >>
> >> ------------------------------
> >> Date: Sat, 30 May 2015 09:15:19 -0300
> >> From:
>
> > alexandre.loss@
>
> >> To:
>
> > mkgmap-dev at .org
>
> >> Subject: Re: [mkgmap-dev] Splitter problem: losing a track segment
> >>
> >>
> >> Hi Gerd,
> >>
> >> Overlap of 10000 fixed my problem. :-)
> >>
> >> Thank you very much!
> >>
> >> Alexandre
> >>
> >> 2015-05-30 8:20 GMT-03:00 Alexandre Loss <
>
> > alexandre.loss@
>
> > >:
> >>
> >> Hi Gerd,
> >>
> >> Thanks for your answer.
> >> I'll try your suggestion and give you a feedback about the result.
> >>
> >> Regards,
> >>
> >> Alexandre
> >>
> >> 2015-05-30 5:43 GMT-03:00 Gerd Petermann <
>
> > gpetermann_muenchen@
>
> > >
> >> :
> >>
> >> Hi Alexandre,
> >>
> >> I don't have much time to analyse your data today.
> >> I think splitter assumes that the tile boundaries
> >> are
> >> "multiples of 0x200 map units wide and high"
> >> when you use resolution=15.
> >>
> >> Since you use keep-complete=false you have to watch
> >> the overlap value.
> >> You can use a larger overlap value like 10000 to fix this problem.
> >>
> >> Gerd
> >>
> >>
> >> ------------------------------
> >> Date: Fri, 29 May 2015 22:04:34 -0300
> >> From:
>
> > alexandre.loss@
>
> >> To:
>
> > mkgmap-dev at .org
>
> >> Subject: [mkgmap-dev] Splitter problem: losing a track segment
> >>
> >>
> >> Hi guys,
> >>
> >> I'm facing a problem in one of my maps, that is probably might be
> >> happening in other places where I didn't see yet.
> >> The problem can be seen in the figure below taken from MapSource. There
> >> is
> >> a gap in a track on the border of the tiles:
> >>
> >> [image: Imagem inline 2]
> >>
> >> So I started to analyze the problem step-by-step till find that the
> >> problem was generated by Splitter (last version).
> >>
> >> To simplify your analysis, I edited my .osm file before running Splitter
> >> and remove all objects (nodes, lines, poi, etc.) not related to the
> >> problem, leaving only the line where the problem happens. The result was
> >> the file "93352150-Ribeirao_Preto.osm" attached, containing only this:
> >>
> >> <?xml version="1.0" encoding="UTF-8"?>
> >>
> > <osm version="0.6" generator="TSuite 6.6.0.4">
> >>
>
> >>
> > <node id="576" lat="-21.9388623" lon="-48.3347007" visible="true"
> >>
> >  version="1"/>
> >>
> > <node id="805" lat="-21.8877920" lon="-48.2932640" visible="true"
> >>
> >  version="1"/>
> >>
> > <node id="806" lat="-21.8674906" lon="-48.2767485" visible="true"
> >>
> >  version="1"/>
> >>
> > <node id="804" lat="-21.8661147" lon="-48.2756044" visible="true"
> >>
> >  version="1"/>
> >>
> > <way id="803" visible="true" version="1">
> >>
> > <nd ref="576"/>
> >>
> > <nd ref="805"/>
> >>
> > <nd ref="806"/>
> >>
> > <nd ref="804"/>
> >>
> > <tag k="highway" v="trunk"/>
> >>
> > <tag k="name" v="SP-255 DANIELLE"/>
> >>
> > <tag k="ref" v="SP-255"/>
> >>
> > <tag k="addr:city" v="BOA ESPERANÇA DO SUL"/>
> >>
> > <tag k="addr:state" v="SP"/>
> >>
> > <tag k="addr:country" v="Brasil"/>
> >>
> > </way>
> >>
>
> >>
> > </osm>
> >>
> >> Then I ran Splitter on this file and the result was the file
> >> "93352150.osm" (attached and compacted), whose content are the lines
> >> bellow:
> >>
> >> <?xml version='1.0' encoding='UTF-8'?>
> >>
> > <osm version='0.5' generator='splitter' upload='false'>
> >>
> > <bounds minlat='-21.895751953125' minlon='-50.262451171875'
> >>
> >  maxlat='-20.401611328125' maxlon='-47.548828125'/>
> >>
> > <node id='805' lat='-21.8877920' lon='-48.2932640'/>
> >>
> > <node id='806' lat='-21.8674906' lon='-48.2767485'/>
> >>
> > <node id='804' lat='-21.8661147' lon='-48.2756044'/>
> >>
> > <way id='803'>
> >>
> > <nd ref='576'/>
> >>
> > <nd ref='805'/>
> >>
> > <nd ref='806'/>
> >>
> > <nd ref='804'/>
> >>
> > <tag k='highway' v='trunk'/>
> >>
> > <tag k='name' v='SP-255 DANIELLE'/>
> >>
> > <tag k='ref' v='SP-255'/>
> >>
> > <tag k='addr:city' v='BOA ESPERANÇA DO SUL'/>
> >>
> > <tag k='addr:state' v='SP'/>
> >>
> > <tag k='addr:country' v='Brasil'/>
> >>
> > </way>
> >>
> > </osm>
> >>
> >> Note that in the "93352150.osm" file is missing the node 576 definition
> >> (i.e.:
> > <node id="576" lat="-21.9388623" lon="-48.3347007" visible="true"
> >>
> >  version="1"/>).
> >> And this is the cause of the lost segment.
> >>
> >> I'm running Splitter in the following way:
> >>
> >> java -jar splitter.jar --split-file=93352150-Ribeirao_Preto.list
> >> --max-areas=1024 --mixed=true --keep-complete=false --resolution=15
> >> --output=xml 93352150-Ribeirao_Preto.osm
> >>
> >> The above .list file is attached.
> >>
> >> If this is not a Splitter error, please, let me know what I'm doing
> >> wrong.
> >>
> >> Thanks.
> >>
> >> Alexandre
> >>
> >>
> >> _______________________________________________ mkgmap-dev mailing list
> >>
>
> > mkgmap-dev at .org
>
> >> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >>
> >> _______________________________________________
> >> mkgmap-dev mailing list
> >>
>
> > mkgmap-dev at .org
>
> >> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >>
> >>
> >>
> >>
> >> _______________________________________________ mkgmap-dev mailing list
> >>
>
> > mkgmap-dev at .org
>
> >> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >>
> >> _______________________________________________
> >> mkgmap-dev mailing list
> >>
>
> > mkgmap-dev at .org
>
> >> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >>
> >
> > _______________________________________________
> > mkgmap-dev mailing list
>
> > mkgmap-dev at .org
>
> > http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >
> > image.png (12K)
> > <http://gis.19327.n5.nabble.com/attachment/5846805/0/image.png>
>
>
>
>
>
> --
> View this message in context:
> http://gis.19327.n5.nabble.com/Splitter-problem-losing-a-track-segment-tp5846557p5846806.html
> Sent from the Mkgmap Development mailing list archive at Nabble.com.
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20150602/7be86300/attachment-0001.html>


More information about the mkgmap-dev mailing list