[mkgmap-dev] Bug in Road Merging - actually doubling roads.

Sat Mar 10 17:29:38 GMT 2018

well we already have code (it's a patch still I think?) to remove roads
with connection to only one side. --> set
mkgmap:set_semi_connected_type=none;
Maybe we could have a filter to remove all semi_connected_type lines with
min length? 4-5m long roads that are only semi connected will be just
adding clutter anyway.

I don't think we lose any essential streets by removing them. Maybe only do
not remove them if an address is attached to them (not autogenerated
address - but fully entered in OSM)?
And yes removing supershort roads and copying the restrictions to the
adjoining ones would be best I guess.

On 10 March 2018 at 18:11, Mike Baggaley <mike at tvage.co.uk> wrote:

> Hi Gerd,
>
> If we take for example that the connection between each two nodes is a
> separate way, with different properties which may affect routing, then any
> properties in the ways in the spike will have no effect on routing. Let's
> say that the side road in the screenshots has no access restrictions
> between
> the 1st and 2nd points, bars traffic between the 2nd and 3rd, then has no
> restrictions after that. With the current code, the access restriction will
> be in the spike and hence will not prevent routing past the end of the
> spike, providing the route doesn't attempt to go down the spike. Based on
> that, as a first stab, we would be no worse off by just deleting the spike
> whatever its access restrictions may be.
>
> The ideal would be that having detected that a road with no junctions has
> looped back to the same co-ordinate, check whether the loopback has access
> restrictions or highway type which are more restrictive than the ones
> either
> side. If not, just delete the spike. If the loopback is more restrictive,
> delete the spike and move the restrictions to one side or the other.
> Probably would also be useful to output a message saying a spike has been
> deleted; the mapper could then review the OSM data with a view to making it
> more mkgmap friendly, providing this can be achieved without compromising
> the OSM representation (I have no idea how many spikes might be detected).
>
> Cheers,
> Mike
>
> -----Original Message-----
> From: Gerd Petermann [mailto:gpetermann_muenchen at hotmail.com]
> Sent: 09 March 2018 16:12
> To: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
> Subject: Re: [mkgmap-dev] Bug in Road Merging - actually doubling roads.
>
> Hi Mike,
>
> the problem in the sample is that there are 5 ways, each with 2 nodes.
> So, we cannot delete any node without also deleting the corresponding way,
> and that would mean that we might remove important routing information.
>
> Besides that you are right, we have to detect that spike. I just think that
> the solution
> is to move all nodes to the same Garmin point, so we have two ways at that
> single point.
>
> Anyway, I just found a stupid bug when handling two points rounded to the
> same Garmin
> point and that bug causes some of the poor results. In line 1213 it should
> be
> if (c.getDistToDisplayedPoint() >= n.getDistToDisplayedPoint())
> instead of
> if (c.getDistToDisplayedPoint() < n.getDistToDisplayedPoint())
>
> I am now trying to find out what other code might be removed when this
> error
> is fixed...
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von
> Mike
> Baggaley <mike at tvage.co.uk>
> Gesendet: Freitag, 9. März 2018 17:01:27
> An: 'Development list for mkgmap'
> Betreff: Re: [mkgmap-dev] Bug in Road Merging - actually doubling roads.
>
> Hi Gerd, I don't know anything about this bit of code, but the problem in
> the before and after pictures look to me like they are not the result of
> attempting to fix the angle, but look like they are the result of the
> second
> and fourth nodes in the spur both rounding to the same point, and it seems
> to me that the solution is to detect that the way has gone back on itself,
> and delete the intermediate node. If there are no junctions between two
> points that loop back on themselves, they should be able to be safely
> deleted before they are passed to WAF. Of course, I may be completely
> misunderstanding the intricacies of the problem, in which case I apologise
> for interrupting.
>
> Cheers,
> Mike
>
> -----Original Message-----
> From: Gerd Petermann [mailto:gpetermann_muenchen at hotmail.com]
> Sent: 09 March 2018 08:16
> To: Development list for mkgmap <mkgmap-dev at lists.mkgmap.org.uk>
> Subject: Re: [mkgmap-dev] Bug in Road Merging - actually doubling roads.
>
> Hi Felix,
>
> okay, thanks.
> BTW: I still have no good idea how to solve the problem with the spike at
> node 1270407079.
> The attached screen shots show the problem. The coloured lines are the
> input.
> Many very short ways and all points are far from the Garmin grid points.
> The picture before.png shows the input to WrongAngleFixer (WAF), the
> after.png the output.
> The output is a bit better, but no improvement regarding the spike.
> I think WAF should collapse the two ways (111528849 and 111528851) which
> build the spike.
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von
> Felix Hartmann <extremecarver at gmail.com>
> Gesendet: Donnerstag, 8. März 2018 20:47:39
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] Bug in Road Merging - actually doubling roads.
>
> Thanks Gerd, I will do some tests on Saturday and then report back.
>
>
> Felix
>
> On 8 March 2018 at 11:46, Gerd Petermann
> <gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com>>
> wrote:
> Hi Felix,
>
> attached is a patch that improves WrongAngleFixer.
> When both ways build a junction and both have another point close to that
> junction
> and in OSM they are different nodes mkgmap tries harder to avoid that they
> are rounded to the
> same coordinate. This is what happened with these two nodes
> https://www.openstreetmap.org/node/1263658804
> https://www.openstreetmap.org/node/1195812701
>
> I found similar situations at some roundabouts, sometimes a way connected
> to
> the roundabout and the roundabout way
> overlap.
> I am not sure how this patch works with your style, please try it with
> r4131.
>
> A few special cases to look at:
> https://www.openstreetmap.org/way/208976154  Coordinates: 38.4082747,
> 16.0787201
> https://www.openstreetmap.org/way/312642720   Coordinates: 39.0259441,
> 8.5967561
> https://www.openstreetmap.org/way/148634420    Coordinates: 41.981263,
> 12.4114831
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev
> <mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-
> dev-bounces at lists.mkgm
> ap.org.uk>> im Auftrag von Felix Hartmann
> <extremecarver at gmail.com<mailto:extremecarver at gmail.com>>
> Gesendet: Samstag, 3. März 2018 19:49:24
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] Bug in Road Merging - actually doubling roads.
>
> well in general 0x16 is not avoided by any transport means (or at least it
> was not a couple of years ago when I checked the influence of types= -
> that's strange. 0x01 and 0x02 are avoided by bicycle - but in general I
> never use avoidances except toll and unpaved - they are pretty straight
> forward. I should do some checks again if something changed with activity
> routing over the years.
> If 6 ways on top of each other are routable problems are common (but do not
> have to occur - not sure why).
>
> (oh yeah I forgot to say to add the route=hiking to relations file, I
> assumed that is clear as it's a route and they are always relations).
>
> On 3 March 2018 at 15:27, Gerd Petermann
> <gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> ><mai
> lto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> >>
> > wrote:
> Hi Felix,
>
> okay, found out again why merging before angle fixing doesn't work, but I
> still have no clue why this junction would cause trouble reg. routing
> with a normal mkgmap binary.
>
> Gerd
>
>
> ________________________________________
> Von: mkgmap-dev
> <mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-
> dev-bounces at lists.mkgm
> ap.org.uk><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:
> mkgmap-dev-b
> ounces at lists.mkgmap.org.uk>>> im Auftrag von Gerd Petermann
> <gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> ><mai
> lto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> >>
> >
> Gesendet: Samstag, 3. März 2018 11:34:24
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] Bug in Road Merging - actually doubling roads.
>
> Hi Felix,
>
> I just noted that I got my results reg. the modified default style with a
> modified StyledConverter which does the road merging before the wrong angle
> fixing.
> Something goes very wrong with this patched version.
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev
> <mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-
> dev-bounces at lists.mkgm
> ap.org.uk><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:
> mkgmap-dev-b
> ounces at lists.mkgmap.org.uk>>> im Auftrag von Gerd Petermann
> <gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> ><mai
> lto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> >>
> >
> Gesendet: Samstag, 3. März 2018 11:01:42
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] Bug in Road Merging - actually doubling roads.
>
> Hi Felix,
>
> no, it is just special at this juntion because the other ways have 0x8 or
> 0x13.
> On thursday I wrote that I was able to reproduce a problem at this junction
> but I forgot to set up routing
> for car (automotive) :-(
> I can reproduce a problem with "tour cycling", but I think that is expected
> with your style,
> because it avoids main roads by default. Maybe your users made the same
> error?
>
> You suggested to add
> route=hiking [0x13 road_class=4 road_speed=2 resolution 24 continue]
> to reproduce a problem. I tried that with no effect. Only after I added a
> rule in the relations file:
> type=route & route=hiking {
>     apply {
>         add route=hiking
>     }
> }
> With that I see additional routable lines in the map and routing is more
> less broken at node 412440328 when
> I try to route from north to anywhere below the node , also the inverse way
> doesn't work.
> I tried different variants of the rule like other types or other road
> classes, always saw the same problem.
> I see working routing again when I remove way 130665733 from the input file
> and move node 1263658617
> down to 412440328 and merge them (see attached file).
> I'll try to find out why this is so special in this situation.
>
> ciao,
> Gerd
>
>
>
>
>
>
>
> ________________________________________
> Von: mkgmap-dev
> <mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-
> dev-bounces at lists.mkgm
> ap.org.uk><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:
> mkgmap-dev-b
> ounces at lists.mkgmap.org.uk>>> im Auftrag von Felix Hartmann
> <extremecarver at gmail.com<mailto:extremecarver at gmail.com><mailto:
> extremecarve
> r at gmail.com<mailto:extremecarver at gmail.com>>>
> Gesendet: Freitag, 2. März 2018 22:21:35
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] Bug in Road Merging - actually doubling roads.
>
> Well in general 0x16 does not cause any problems. Do you think if I use
> another type instead (e.g.) 0x13 it won't be problematic, or do you mean
> that the 0x16 in itself is the problem?
>
> On 1 March 2018 at 18:55, Gerd Petermann
> <gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> ><mai
> lto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> >>
> <mailto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_
> muenchen at hotmail.c
> om><mailto:gpetermann_muenchen at hotmail.com<mailto:gp
> etermann_muenchen at hotmai
> l.com>>>> wrote:
> Hi Felix,
>
> okay, I think I can reproduce a routing problem on my Oregon with your map.
> I have a route that works from south to north but not in the reverse
> direction.
> I'll try to create a similar situation with a modified default style to
> find
> out
> what the problem is. My current prime suspect is the way with type 0x16.
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev
> <mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-
> dev-bounces at lists.mkgm
> ap.org.uk><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:
> mkgmap-dev-b
> ounces at lists.mkgmap.org.uk>><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk
> <m
> ailto:mkgmap-dev-bounces at lists.mkgmap.org.uk><mailto:mk
> gmap-dev-bounces at list
> s.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk>>>> im
> Auftrag
> von Felix Hartmann
> <extremecarver at gmail.com<mailto:extremecarver at gmail.com><mailto:
> extremecarve
> r at gmail.com<mailto:extremecarver at gmail.com>><mailto:extremec
> arver at gmail.com<
> mailto:extremecarver at gmail.com><mailto:extremecarver at gmail.com<mailto:
> extrem
> ecarver at gmail.com>>>>
> Gesendet: Donnerstag, 1. März 2018 18:09:02
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] Bug in Road Merging - actually doubling roads.
>
> routing issues is on devices only - not in Basecamp. Reported so far from
> Oregon and Edge - routing from North to East over that road.
>
>
> Well what I thought was that mkgmap should check if another road is put
> above. Right now it's 1 road supposed to be there, 2 and 3 roads
> overlapping
> due to that road being too short. Those overlapping somehow would need to
> be
> removed (and yes I know they are originally not the same line). My style
> should I think in maximum create 4 routable lines - however due to
> overlapping of other lines (due to the sub 4.3m line) it ends up a short
> stretch with more than 5 routable lines.
>
> On 1 March 2018 at 17:47, Gerd Petermann
> <gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> ><mai
> lto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com
> >>
> <mailto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_
> muenchen at hotmail.c
> om><mailto:gpetermann_muenchen at hotmail.com<mailto:gp
> etermann_muenchen at hotmai
> l.com>>><mailto:gpetermann_muenchen at hotmail.com<mailto:gp
> etermann_muenchen at h
> otmail.com><mailto:gpetermann_muenchen at hotmail.com<mailto:gp
> etermann_muenche
> n at hotmail.com>><mailto:gpetermann_muenchen at hotmail.com<mailto:
> gpetermann_mue
> nchen at hotmail.com><mailto:gpetermann_muenchen at hotmail.com<mailto:
> gpetermann_
> muenchen at hotmail.com>>>>> wrote:
> Hi Felix,
>
> Felix Hartmann-2 wrote
> > Okay - no-merge-lines makes no difference, also removing
> > addArcsToMajorRoads();
> > function makes no difference. And I just noticed - the patch does not
> > work.
> > It works in it's own way - but not achieving the result that is needed.
> > Whatever I do - I still end up with 6 routable lines.
> >
> > I guess the patch checks for each road that it is not more than 4 times
> in
> > the map. So I end up with 1, 2, and 3 (1 that is supposed to be in the
> > map,
> > 2 from the east not supposed, and 3 from the south, not supposed).
> (always
> > using gpsmapedit to check results).
>
> yes, that is what it does. Didn't you suggest something like that?
>
> You still did not tell me how I can reproduce a crash with this data, so I
> can only guess.
> I didn't see any routing issues so far with your maps, only the ugly
> rendering.
> Is this the only problem or is there really a routing issue?
> If yes, please let me know how to reproduce it.
>
> Gerd
>
>
>
>
> --
> Sent from: http://gis.19327.n8.nabble.com/Mkgmap-Development-f5324443.html
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk
> ><mailto
> :mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk
> >><mail
> to:mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk
> ><mai
> lto:mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk
> >>><
> mailto:mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-
> dev at lists.mkgmap.org.uk>
> <mailto:mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-
> dev at lists.mkgmap.org.uk
> >><mailto:mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-
> dev at lists.mkgmap.org.
> uk><mailto:mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-
> dev at lists.mkgmap.org
> .uk>>>>
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>
>
>
> --
> Felix Hartman - Openmtbmap.org & VeloMap.org
> Schusterbergweg 32/8
> 6020 Innsbruck
> Austria - Österreich
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk
> ><mailto
> :mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk
> >><mail
> to:mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk
> ><mai
> lto:mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk
> >>>
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>
>
>
> --
> Felix Hartman - Openmtbmap.org & VeloMap.org
> Schusterbergweg 32/8
> 6020 Innsbruck
> Austria - Österreich
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk
> ><mailto
> :mkgmap-dev at lists.mkgmap.org.uk<mailto: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<mailto:mkgmap-dev at lists.mkgmap.org.uk
> ><mailto
> :mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk>>
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>
>
>
> --
> Felix Hartman - Openmtbmap.org & VeloMap.org
> Schusterbergweg 32/8
> 6020 Innsbruck
> Austria - Österreich
>
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk>
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>
>
>
> --
> Felix Hartman - Openmtbmap.org & VeloMap.org
> Schusterbergweg 32/8
> 6020 Innsbruck
> Austria - Österreich
>
> _______________________________________________
> 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
>

-- 
Felix Hartman - Openmtbmap.org & VeloMap.org
Schusterbergweg 32/8
6020 Innsbruck
Austria - Österreich
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20180310/5be107aa/attachment-0001.html>