[mkgmap-dev] RE Commit r3801: merge split-shape branch

Tue Feb 21 09:02:48 GMT 2017

Hi Ticker,

I thought about this for a while and I am still not sure where to go.

One problem with the high-prec stuff is that it is missleading.
When I started to code the highprec() code my only intention was to be able to
retreive "the original" position as coded in OSM.
You probably learned it the hard way that the value returned by Coord.getLatitude()
sometimes is different to the value from a rounded Coord.getHighPrecLat() .
The reason is that functions like Coord.getAlternativePositions() used by WrongAngleFixer
create Coord instances with the same highprec positions but different rounding (a bit more to
the left or right, a bit more up or down). The idea is that the map looks better when
the error for that single point is increased. A good place to check this are multiple parallel rails.

I am still trying to find good code to calculate the bounding box for coords with this special case.
When such a point is close to or "on" the bounding box in one precision it might be outside
with the other. I wonder if it would be better to use low precision as soon as possible.

If you have a good idea for that please let me know.

In the meantime I have found some reasons to change precision from 30 to 32 bit as this helps when
checking mp-relations. I found new problems, e.g. 180.0 degrees and -180.0 degrees started to give the
same 32 bit value. Working on that now...

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, 17. Februar 2017 15:49:10
An: mkgmap-dev at lists.mkgmap.org.uk
Betreff: Re: [mkgmap-dev] RE Commit r3801: merge split-shape branch

Hi Gerd

Thoughts were:

I was having problems with a multi-polygon relation where OSM showed a
very small gap between 2 parts, but some final cutting of the shape
found that these lines now intersected, and I thought it could be a
mapPrec vs highPrec problem where the multiPolygonRelation logic is
looking for this sort of thing. Changing it to highPrec didn't fix it
(you found that the problem was roundCoord filter) but the changes
looked worth keeping because:

MultiPolygon cutting to expose holes uses highPrec and it seemed best
to be consistent, avoiding cuts that might leave very small areas.

A general move in the direction of all coordinates/bounds/etc being the
held only in highPrec and resolution 24 not being a special case.

But I don't have strong feelings about it.

Ticker

On Fri, 2017-02-17 at 13:12 +0000, Gerd Petermann wrote:
> Hi Ticker,
>
> I did not see an advantage in using high-prec. Why do you think it
> should be used?
>
> 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, 17. Februar 2017 13:27:28
> An: mkgmap-dev at lists.mkgmap.org.uk
> Betreff: Re: [mkgmap-dev] RE Commit r3801: merge split-shape branch
>
> Hi Gerd
>
> Yes - sorry, my fault.
> I've just tested a fix to this and the sea is OK now. Is is worth
> keeping the high-res changes in the split-shape branch pending more
> work on MultiPolygonRelation? If so I'll commit.
>
> Ticker
>
>
> On Fri, 2017-02-17 at 10:49 +0000, Gerd Petermann wrote:
> > Hi Ticker,
> >
> > problem is in these lines:
> >                         if (remove) {
> >                                 // check if the polygon contains
> > the
> > complete bounding box
> >                                 if
> > (w.getBounds().contains(tileArea.getBounds())) {
> >                                         remove = false;
> >                                 }
> >                         }
> > 1) The SeaGenerator creates an outer way which is larger than the
> > tile bounds.
> > 2) This test was wrong because w.getBounds() returns a bbox  in
> > highprec values and tileArea.getBounds()  is in Garmin units
> >
> > 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, 17. Februar 2017 11:00:49
> > An: mkgmap-dev at lists.mkgmap.org.uk
> > Betreff: Re: [mkgmap-dev] RE Commit r3801: merge split-shape branch
> >
> > Hi Gerd
> >
> > Good.
> >
> > Ticker
> >
> > On Fri, 2017-02-17 at 09:41 +0000, Gerd Petermann wrote:
> > > Hi Ticker,
> > >
> > > attached is the patch I've created with help of svn.
> > > It only reverts the changes made to MultiPolygonRelation in
> > > r3771.
> > >
> > > I don't know yet why they cause trouble.
> > >
> > > 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, 17. Februar 2017 10:32:11
> > > An: mkgmap-dev at lists.mkgmap.org.uk
> > > Betreff: Re: [mkgmap-dev] RE Commit r3801: merge split-shape
> > > branch
> > >
> > > Hi Gerd
> > >
> > > Yes - if that fixes it.
> > > But
> > >  in r3771 were there a couple of other changes not to
> > > MultiPolygon...
> > >  haven't a lot more changes been made since to MultiPolygon...
> > >
> > > Ticker
> > >
> > >
> > > On Fri, 2017-02-17 at 09:19 +0000, Gerd Petermann wrote:
> > > > Hi Ticker,
> > > >
> > > > I think the problems were introduced with the highprec changes
> > > > in
> > > > MultiPolygonRelation in r3771.
> > > > When I revert those the problem is gone. If you don't mind I
> > > > commit
> > > > this revert.
> > > >
> > > > Gerd
> > > > ________________________________________
> > > > Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im
> > > > Auftrag
> > > > von Gerd Petermann <GPetermann_muenchen at hotmail.com>
> > > > Gesendet: Freitag, 17. Februar 2017 10:02:38
> > > > An: Development list for mkgmap
> > > > Betreff: Re: [mkgmap-dev] RE Commit r3801: merge split-shape
> > > > branch
> > > >
> > > > Hi Ticker,
> > > >
> > > > don't need test data, found out what you meant.
> > > >
> > > > Gerd
> > > > ________________________________________
> > > > Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im
> > > > Auftrag
> > > > von Gerd Petermann <GPetermann_muenchen at hotmail.com>
> > > > Gesendet: Freitag, 17. Februar 2017 09:52:56
> > > > An: Development list for mkgmap
> > > > Betreff: Re: [mkgmap-dev] RE Commit r3801: merge split-shape
> > > > branch
> > > >
> > > > Hi Ticker,
> > > >
> > > > okay, I'll merge the branch into trunk.
> > > > Please post some test data to show the coastline problem and
> > > > I'll
> > > > try
> > > > to fix it.
> > > >
> > > > 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, 17. Februar 2017 09:50:17
> > > > An: mkgmap-dev at lists.mkgmap.org.uk
> > > > Betreff: Re: [mkgmap-dev] RE Commit r3801: merge split-shape
> > > > branch
> > > >
> > > > Hi Gerd
> > > >
> > > > I fixed this problem with r3807 in the split-shape branch and
> > > > this
> > > > should probably be merged into trunk.
> > > >
> > > > Mike was also getting another problem that might be related to
> > > > generate-sea:extend-sea-sectors
> > > >
> > > > I don't know much about this area, but just having this option
> > > > and
> > > > nothing else of significance, it seems as if the overview map
> > > > has
> > > > all
> > > > sea if it has a coastline
> > > >
> > > > Ticker
> > > >
> > > > On Thu, 2017-02-16 at 20:12 +0000, Gerd Petermann wrote:
> > > > > Hi Ticker,
> > > > >
> > > > > In case you need test data: I've uploaded a file :
> > > > > http://files.mkgmap.org.uk/download/334/88009211.osm.pbf
> > > > > Produces the error with r3811 without any options, just
> > > > > java -jar mkpgmap.jar 88009211.osm.pbf
> > > > > Seems to work okay with r3807.
> > > > >
> > > > > Gerd
> > > > >
> > > > > ________________________________________
> > > > > Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im
> > > > > Auftrag
> > > > > von Mike Baggaley <mike at tvage.co.uk>
> > > > > Gesendet: Donnerstag, 16. Februar 2017 10:40:55
> > > > > An: mkgmap-dev at lists.mkgmap.org.uk
> > > > > Betreff: Re: [mkgmap-dev] RE Commit r3801: merge split-shape
> > > > > branch
> > > > >
> > > > > HI Ticker, I'll package up some data for you. I have generate
> > > > > -sea:extend-sea-sectors in my options file, I think it will
> > > > > be
> > > > > related to that.
> > > > >
> > > > > Regards,
> > > > > Mike
> > > > >
> > > > > -----Original Message-----
> > > > > From: Ticker Berkin [mailto:rwb-mkgmap at jagit.co.uk]
> > > > > Sent: 16 February 2017 08:17
> > > > > To: mkgmap-dev at lists.mkgmap.org.uk
> > > > > Subject: Re: [mkgmap-dev] RE Commit r3801: merge split-shape
> > > > > branch
> > > > >
> > > > > Hi Mike
> > > > >
> > > > > I'll fix/improve the message.
> > > > >
> > > > > Can you tell if it is the sea polygons, the line 'coastline',
> > > > > or
> > > > > the
> > > > > island area (if your style generates this) that is corrupt.
> > > > >
> > > > > Maybe send your splitter options / areas.list, mkgmap options
> > > > > and
> > > > > style
> > > > > and I'll see if I can reproduce.
> > > > >
> > > > > Ticker
> > > > >
> > > > > On Thu, 2017-02-16 at 01:23 +0000, Mike Baggaley wrote:
> > > > > > HI Ticker, I've tried the patch without the --order-by
> > > > > > -decreasing
> > > > > > -area option and it does now run without crashing. However,
> > > > > > it
> > > > > > suffers from the same problem as I mentioned earlier when
> > > > > > trying
> > > > > > -
> > > > > > -order-by-decreasing-area without the patch, in that the
> > > > > > coastline
> > > > > > is
> > > > > > corrupted. I haven't yet turned on detailed logging - I can
> > > > > > do
> > > > > > that
> > > > > > if it would help.
> > > > > >
> > > > > > I note that in MapSplitter.java there is a new line:
> > > > > >
> > > > > > log.info("Single item larger that WANTED_MAX_AREA_SIZE",
> > > > > > area.getBounds().getCenter().toOSMURL());
> > > > > >
> > > > > > I think this should say 'than' rather than 'that' and
> > > > > > probably
> > > > > > WANTED_MAX_AREA_SIZE should be outside the string so we see
> > > > > > an
> > > > > > actual
> > > > > > number.
> > > > > >
> > > > > > Regards,
> > > > > > Mike
> > > > >
> > > > >
> > > > >
> > > > > _______________________________________________
> > > > > 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
> > > _______________________________________________
> > > 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