[mkgmap-dev] Commit 4710

Thu May 13 14:14:33 BST 2021

Hi all,

I fear I've totally forgotten the case of extended line types. I think the current code doesn't write the direction flag for them and I don't know if they can have a direction.

Gerd

________________________________________
Von: Gerd Petermann <gpetermann_muenchen at hotmail.com>
Gesendet: Donnerstag, 13. Mai 2021 14:33
An: Development list for mkgmap
Betreff: AW: [mkgmap-dev] Commit 4710

Hi Felix,

are you sure that you tested the two versions with exactly the same input? (osm data, style, options)?
If the changes in r4710 or r4711 really cause differences in routing quality there must be an error, either in my understanding or in the code.
Maybe you can post links to two tiles where routing differs?

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Felix Hartmann <extremecarver at gmail.com>
Gesendet: Donnerstag, 13. Mai 2021 14:24
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] Commit 4710

Yes, 4711 on branch vs best version 4709 on branch. 4709 was best so far.

On Thu, 13 May 2021, 19:37 Gerd Petermann <gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com>> wrote:
Hi Felix,

you totally lost me. There is no version r4713 in the branch.
It seems you report the version number that svn shows after an svn update? That's not relevant, you must use svn info to find out the version of your branch.
svn update always shows the latest commit, no matter in what branch it was made.

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: Donnerstag, 13. Mai 2021 13:16
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] Commit 4710

I just looked it up. It must have been 4709 with best routing for me (unlikely but maybe it could have been 4708), while 4711 is a bit worse (but better than before the first changes that made an impact on routing). Both from low-res-opt branch.
I haven't tried trunk for quite a while.

On Thu, 13 May 2021 at 17:42, 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 Felix,

please tell me exactly which versions you tested reg. routing.
Note that the branches do not yet contain the latest changes in trunk.

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk><mailto: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><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com>>>
Gesendet: Donnerstag, 13. Mai 2021 11:28
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] Commit 4710

Well I just got to test routing - and the current version is a degradation vs the intermediate version from yesterday for my maps. The current version routes better than before, but worse than the intermediate version.

As for the list - it is a bit more complicated inside the style but doable. I really do not know what happens to those ways where I first set a direction for an downhill only way with higher priority, then remove the direction for a way that can be used in both directions at lower priority. Maybe sometimes also doing this the other way around. A list with type and max resolution used would be much easier for writing your style. Instead of adding 60-100 lines with the filter it would be done with 10 or so.  Canals should not have a direction, they do not flow. Sidewalk=left, right is the same as for the various cylceway,cycletrack options - those are really not reversible. The underlying way could be reversed however - and I guess they are only used for level 0.
Oneway arrows for streets are maybe used from resoltuion 24-22 or 24 only. Cliffs are also only high resolution. For me rivers are the only thing which really go into lower resolutions.

On Thu, 13 May 2021 at 16:03, 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>><mailto: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 Felix,

FYI:
I just found out that the current code to detect if the style sets mkgmap:has-direction=true doesn't work. Seems I didn't test this :(
The change in r4710 changed only the LineMerger in the branch. Even with r4711 LineMerger only merges roads in maps without NET, at least that's what's intended.

My understanding is that r4703 changed routing, possibly also r4704. The merge from trunk also changed routing in the branches (r4706 and r4707).

I'm now fixing the detection code, next I'm trying to figure out how to configure the details reg. direction handling.

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk>><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk><mailto: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><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com>><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com>>>>
Gesendet: Donnerstag, 13. Mai 2021 09:52
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] Commit 4710

I kinda feel by default even oneway=yes should only mean do not change direction for level 0. If someone uses a style to have arrows showing the oneway, then only for that arrow line (defined by tag) the direction cannot be reversed. Yes the problem of DP filter rests - I feel this does not matter for resolution 24 and even 23, but if you display arrows for resolution 22 or higher then a tag should not merge the underlying lines. Besides rivers (then also many styles do not have an arrow for rivers, or you could decide to show the arrows only at resolution 24 and 23) few lines will be objection dependent.

I will try out now if there is any change in routing - but I will only write again if there unexpectedly is. The sharp angles changed routing for the better for my maps by quite a lot. So maybe with now some less lines being merged, there actually will be a little change for the worse back to the old behavior. Not sure..

On Thu, 13 May 2021 at 14:07, 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>><mailto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com><mailto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com>>><mailto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com><mailto:gpetermann_muenchen at hotmail.com<mailto:gpetermann_muenchen at hotmail.com>><mailto: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 Felix,

yes, size increases if your style sets mkgmap:has-direction=true. I just want to make sure that the direction flag is treated correctly first.
As already discussed we might introduce a new option or tag to tell mkgmap the min. level at which the direction has to be kept. You suggested to ignore direction at level > 0, I think it might depend on the style and TYP. I'll play with the OFM style to find out more.

The change should not affect routing at all (none of the changes in the low-res-opt branch should).

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk>><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk>>><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk>><mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk<mailto:mkgmap-dev-bounces at lists.mkgmap.org.uk><mailto: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><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com>><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com>>><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com>><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com><mailto:extremecarver at gmail.com<mailto:extremecarver at gmail.com>>>>>
Gesendet: Donnerstag, 13. Mai 2021 02:59
An: Development list for mkgmap
Betreff: [mkgmap-dev] Commit 4710

fix error in LineMergeFilter reg. lines with direction
The line merger should not merge lines if one has the direction flag set and the other has not. Problem exists also in trunk.

Hmm fixing this stopped all the nice size optimization. Map size got much bigger again.
I did not find any place where this mattered. Routing was also not affected badly. Maybe I did not look good enough?

I do not see why not to merge them. As long as it`s not the opposite it seems fine...
Map size increase/decrease is around 1.5% with my style. So thats quite a big difference.
--
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><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<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>>>>
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<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>>>
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<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

_______________________________________________
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