[mkgmap-dev] new branch split-shape

Mon Jan 23 09:50:57 GMT 2017

Hi Gerd

What aspect of sea polygons is getting worse?

Once the new algo is shown not to cause problems: for quick/simple
wins, I'd like to convert PolygonSplitter*, AreaClipper and
MultiPolygonRelation to use the new code.

I agree that ShapeMergeFilter should operate on the full map rather
than within subdivisions. However, with --order-by-decreasing-area, one
of the good things it does is put back together polygons that got split
by an earlier point-limit filter, but, when split again into
subdivisions the limit is no longer exceeded.

As per your ideas list, having a zoom-filtered representation (or at
least the relevant sizes) of the complete poly/lines before subdivision
splitting starts to get round this, along with delaying splitting as
long as possible.

I like the idea of having the holes in polygons kept as their own
entity, rather than the polygon being cut up. Sometimes the hole will
also be an item to output to the map.

MapArea splitting will need a bit of a restructure. It needs to have a
firmer idea of when division by geographic area has become pointless
(area too small or one half is empty) then switch to another method

Ticker

On Mon, 2017-01-23 at 08:10 +0000, Gerd Petermann wrote:
> Hi Ticker,
> 
> I did some more tests. With --order-by-decreasing-area enabled the
> branch is much faster, in fact
> the additional polygon splits are done with nearly no impact on run
> time, great!
> Still confusing for me is that the option has a negative effect on
> the sea polygons, so I 'll
> have a closer look at this.
> I think we have two possible ways to go from here:
> 1) replace the existing algos with yours to improve performance
> 2) try to implement my idea described on the to do list beginning
> with "rewrite classes that hold"
> http://www.mkgmap.org.uk/dev/todo
> esp. the part regarding "merge shapes before splitting the map into
> sub areas"
> 
> Do you think that this would also work with the --order-by-decreasing
> -area option?
> 
> Gerd
> 
> ________________________________________
> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag
> von Ticker Berkin <rwb-mkgmap at jagit.co.uk>
> Gesendet: Sonntag, 22. Januar 2017 19:23:53
> An: mkgmap-dev at lists.mkgmap.org.uk
> Betreff: Re: [mkgmap-dev] new branch split-shape
> 
> Hi Gerd
> 
> I've merged recent trunk changes into the branch and committed it.
> 
> Ticker
> 
> On Fri, 2017-01-20 at 18:17 +0000, Gerd Petermann wrote:
> > Hi Ticker,
> > 
> > yes, please commit it so that others can test this version.
> > 
> > Gerd
> > 
> > ________________________________________
> > Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag
> > von Ticker Berkin <rwb-mkgmap at jagit.co.uk>
> > Gesendet: Freitag, 20. Januar 2017 16:54:16
> > An: mkgmap-dev at lists.mkgmap.org.uk
> > Betreff: Re: [mkgmap-dev] new branch split-shape
> > 
> > Hi Gerd
> > 
> > I've fixed ShapeSplitter and committed it to branches/split-shape/
> > 
> > It runs through:
> >     3094949146 Jan 13 10:00 germany-latest.osm.pbf
> > with your areas.list without problems. However it is quite
> > sensitive
> > to
> > self-intersecting polygons and these might generate warnings into
> > mkgmap.log.0 if enabled.
> > 
> > I haven't committed my merge of the trunk into this branch, but can
> > if
> > you want.
> > 
> > Ticker
> > 
> > 
> > On Fri, 2017-01-13 at 08:56 +0000, Gerd Petermann wrote:
> > > Hi Ticker,
> > > 
> > > here is my areas.list for splitter (use with --split
> > > -file=areas.list)
> > > 
> > > Gerd
> > > 
> > > ________________________________________
> > > Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im
> > > Auftrag
> > > von Ticker Berkin <rwb-mkgmap at jagit.co.uk>
> > > Gesendet: Freitag, 13. Januar 2017 09:54:01
> > > An: mkgmap-dev at lists.mkgmap.org.uk
> > > Betreff: Re: [mkgmap-dev] new branch split-shape
> > > 
> > > Hi Gerd
> > > 
> > > I thought this case would never happen. I think I know how to
> > > make
> > > it
> > > work. I'll download .de from geofabik and try my std split and
> > > params
> > > to see if I can reproduce. If not I'll ask you for input
> > > files/split
> > > params/cfg...
> > > 
> > > Ticker
> > > 
> > > On Thu, 2017-01-12 at 23:10 -0700, Gerd Petermann wrote:
> > > > Hi Ticker,
> > > > 
> > > > Ticker Berkin wrote
> > > > > Steve changed some stuff and I've now been able to commit it
> > > > > to
> > > > > the
> > > > > branch.
> > > > 
> > > > I tested with a map for germany and got some error messages.
> > > > I used this command:
> > > > java -Xmx6G -jar d:\mkgmap-split-shape\dist\mkgmap.jar -c
> > > > c:\temp\split\mkgmap.cfg --order-by-decreasing-area -c
> > > > c:\osm_out_work\germany\20170111_172616\template.args
> > > > Attached is the mkgmap.cfg.
> > > > Please let me know when you need the corresponding input files.
> > > > 
> > > > SCHWERWIEGEND (ShapeSplitter):
> > > > c:\osm_out_work\germany\20170111_172616\63240043.o5m: Direction
> > > > of
> > > > shape and
> > > > hole conflict
> > > > http://www.openstreetmap.org/?mlat=53.660850&mlon=9.991379&zoom
> > > > =1
> > > > 7
> > > > SCHWERWIEGEND (ShapeSplitter):
> > > > c:\osm_out_work\germany\20170111_172616\63240046.o5m: Lines hit
> > > > divider at
> > > > same points 26865876 26866066
> > > > http://www.openstreetmap.org/?mlat=53.951411&mlon=9.007551&zoom
> > > > =1
> > > > 7
> > > > SCHWERWIEGEND (ShapeSplitter):
> > > > c:\osm_out_work\germany\20170111_172616\63240089.o5m: Lines hit
> > > > divider at
> > > > same points 20894468 20894665
> > > > http://www.openstreetmap.org/?mlat=52.092962&mlon=7.005482&zoom
> > > > =1
> > > > 7
> > > > SCHWERWIEGEND (ShapeSplitter):
> > > > c:\osm_out_work\germany\20170111_172616\63240110.o5m: Direction
> > > > of
> > > > shape and
> > > > hole conflict
> > > > http://www.openstreetmap.org/?mlat=50.982076&mlon=11.418979&zoo
> > > > m=
> > > > 17
> > > > SCHWERWIEGEND (ShapeSplitter):
> > > > c:\osm_out_work\germany\20170111_172616\63240157.o5m: Direction
> > > > of
> > > > shape and
> > > > hole conflict
> > > > http://www.openstreetmap.org/?mlat=47.570801&mlon=7.906524&zoom
> > > > =1
> > > > 7
> > > > SCHWERWIEGEND (ShapeSplitter):
> > > > c:\osm_out_work\germany\20170111_172616\63240168.o5m: Lines hit
> > > > divider at
> > > > same points 29218589 29218749
> > > > http://www.openstreetmap.org/?mlat=48.099174&mlon=9.796295&zoom
> > > > =1
> > > > 7
> > > > SCHWERWIEGEND (ShapeSplitter):
> > > > c:\osm_out_work\germany\20170111_172616\63240184.o5m: Lines hit
> > > > divider at
> > > > same points 23869233 23869283
> > > > http://www.openstreetmap.org/?mlat=50.964096&mlon=8.002801&zoom
> > > > =1
> > > > 7
> > > > SCHWERWIEGEND (ShapeSplitter):
> > > > c:\osm_out_work\germany\20170111_172616\63240205.o5m: Lines hit
> > > > divider at
> > > > same points 147494300 147494521
> > > > http://www.openstreetmap.org/?mlat=49.451318&mlon=7.788942&zoom
> > > > =1
> > > > 7
> > > > 
> > > > mkgmap.cfg <
> > > > http://gis.19327.n8.nabble.com/file/n5889336/mkgmap.cfg>  ;;;;
> > > > 
> > > > 
> > > > 
> > > > --
> > > > View this message in context:
> > > > http://gis.19327.n8.nabble.com/new-bran
> > > > ch-split-shape-tp5889303p5889336.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
> > > _______________________________________________
> > > mkgmap-dev mailing list
> > > mkgmap-dev at lists.mkgmap.org.uk
> > > http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> > > _______________________________________________
> > > mkgmap-dev mailing list
> > > mkgmap-dev at lists.mkgmap.org.uk
> > > http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> > _______________________________________________
> > mkgmap-dev mailing list
> > mkgmap-dev at lists.mkgmap.org.uk
> > http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> > _______________________________________________
> > mkgmap-dev mailing list
> > mkgmap-dev at lists.mkgmap.org.uk
> > http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev