[mkgmap-dev] patch: improve RoadMerger: allow to merge roads with reversed points

Thu May 6 16:42:04 BST 2021

Could you directly push those patches to the low-res branch? I guess anyone
using it right now would like to try it with the patch..
I will try it out tomorrow.

On Thu, 6 May 2021 at 18:32, Gerd Petermann <gpetermann_muenchen at hotmail.com>
wrote:

> Ahh, sorry, merge-roads-reversed-v2.patch was work-in-progress. v3 works
> as described before.
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von
> Gerd Petermann <gpetermann_muenchen at hotmail.com>
> Gesendet: Donnerstag, 6. Mai 2021 12:30
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] patch: improve RoadMerger: allow to merge roads
> with reversed points
>
> Hi all,
>
> here is the 2nd approach:
>
> RoadMerger can connect more roads if direction of the roads doesn't matter
> reg. routing.  Problem is that mkgmap doesn't know if direction matters
> unless the oneway tag
> says so. The patch
> - implements new special tag mkgmap:has-direction=true which must be set
> for ways which shoud not be reversed.
> I am not yet sure how to use this. My idea was that the style author knows
> which types are direction depended and this should add
> mkgmap:has-direction=true for those lines which should not be reversed.
> Before road merging all map lines which were created for a given OSM way
> are checked. If any has a direction then all lines for the given OSM way
> are marked to have a direction.
> Maybe this needs more thinking.
>
> - The program checks if there is any line rule that uses
> mkgmap:has-direction=true. If not, mkgmap will not reverse roads for better
> merging.
> - implement new experimental option --x-force-reverse-merge to override
> the logic reg. mkgmap:has-direction=true.
>
> Reg. size the patch is probably not worth the effort. It just merges a few
> more roads (I see maybe 3% more merges) and thus safes some KB in NET and
> NOD.
> I see this also as a preparation to do more merging in the low-res-opt
> branch.
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von
> Gerd Petermann <gpetermann_muenchen at hotmail.com>
> Gesendet: Mittwoch, 5. Mai 2021 16:53
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] patch: improve RoadMerger: allow to merge roads
> with reversed points
>
> Hi Felix,
>
> the patch contained two bugs:
> - Roads with different direction are not merged when they are connected at
> the beginning.
> - angle calculation was wrong, maybe that rejected possible merges or did
> the worse ones.
>
> When I talk about overlays I mean those lines which are created with the
> continue statement or with overlays rules.
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von
> Felix Hartmann <extremecarver at gmail.com>
> Gesendet: Mittwoch, 5. Mai 2021 16:46
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] patch: improve RoadMerger: allow to merge roads
> with reversed points
>
> If the direction of a line is changed, will all lines created via continue
> also be changed? If yes then this decreases the lines a lot where I would
> need the special tag. If not it will get complicated...
> Well and the patch itself so far only saved about 0.3% of total map size.
> Better than nothing for no visual impact of course.
>
> On Wed, 5 May 2021 at 22:43, Gerd Petermann <
> gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com>>
> wrote:
> Hi Felix,
>
> no idea why you talk about waterways here. The thread is about a patch for
> RoadMerger. As the name suggest this code merges roads, nothing else.
> RoadMerger considers far more attributes of lines compared to the simple
> LineMerger, not just the type and name but also all the attributes which
> affect routing.
> It will not merge oneways with opposite direction, nor does it merge roads
> which are used in restrictions.
>
> My current approach is to implement the new special
> mkgmap:has-direction=true
> If the style sets this for a road or any of its overlay lines RoadMerger
> should not reverse the road.
>
> Gerd
> ________________________________________
> Von: mkgmap-dev <mkgmap-dev-bounces at lists.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>>
> Gesendet: Mittwoch, 5. Mai 2021 13:04
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] patch: improve RoadMerger: allow to merge roads
> with reversed points
>
> Actually when I think about it - reversing the direction for waterways
> should be done anyhow - there is no waterway that flows in opposite
> direction. Cliff lines most likely the same.
> And for lines that are off center - that doesn't matter as long as all
> resulting lines are reversed. It would actually be better as otherwise at
> one point the right side jumps to the left and this could be prevented.
>
> Yes oneway roads (plus all copies created from it via continue) - those
> you cannot merge.
>
> On Wed, 5 May 2021 at 18:47, Felix Hartmann <extremecarver at gmail.com
> <mailto:extremecarver at gmail.com><mailto:extremecarver at gmail.com<mailto:
> extremecarver at gmail.com>>> wrote:
> will that patch also work on the low resolution branch, or only against
> trunk?
>
> On Wed, 5 May 2021 at 18:05, Gerd Petermann <
> GPetermann_muenchen at hotmail.com<mailto:GPetermann_muenchen at hotmail.com
> ><mailto:GPetermann_muenchen at hotmail.com<mailto:
> GPetermann_muenchen at hotmail.com>>> wrote:
> Hi all,
>
> up to now RoadMerger doesn't merge (combine) two roads if the have
> different directions. This is OK for oneway roads but I see no need for
> this with other roads. The attached patch changes this. It has an influence
> on file size (safes a few KB in each tile) and very likely also on routing,
> hopefully to the better/faster.
> It also reduces zig-zagging in the low-res-opt branch.
>
> Strangely enough almost all was already implemented, so maybe WanMil (the
> original author) tried that before and found a reason not to do it. Will
> search the archives later...
>
> Gerd
> _______________________________________________
> 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>>
> https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>
>
> --
> Felix Hartman - Openmtbmap.org & VeloMap.org
>
>
>
> --
> Felix Hartman - Openmtbmap.org & VeloMap.org
>
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk<mailto:mkgmap-dev at lists.mkgmap.org.uk>
> https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>
>
> --
> Felix Hartman - Openmtbmap.org & VeloMap.org
>
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

-- 
Felix Hartman - Openmtbmap.org & VeloMap.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20210506/991e976b/attachment.html>