Hi Gerd, please try again my sample from yesterday but observe second node: http://www.openstreetmap.org/node/2650495573 Maybe your patch is valid only for first cycleway but then it goes out of sync? -- Best regards, Andrzej

Hi Gerd, it is a bit better with --preserve-element-order but still not a complete solution. I have uploaded bigger sample from Paris: http://files.mkgmap.org.uk/download/186/29467175.osm.pbf Please look at node: http://www.openstreetmap.org/node/2650495625 -- Best regards, Andrzej

Hi Andrzej, forgot to mention that GPSMapEdit shows a lot of routing nodes for the ways were mkgmap created a cycleway. AFAIK these nodes are not in the img file, it seems that GPSMapEdit calculates them itself. Gerd GerdP wrote

Hi Andrzej,

here is version 3 of the patch. I still see small changes in the layout of the ways, but not at points which are not crossings. For the ways that are crossings I assume that the WrongAngleFixer switiches between two non-optimal solutions. Depending on the number of roads (odd,even), it switches to a different solution. I'lll look at that part again when I've finished the new code for route restrictions.

Gerd

...

Date: Sat, 22 Mar 2014 16:44:15 +0100 From:

popej@.onet

...

To:

mkgmap-dev@.org

...

Subject: Re: [mkgmap-dev] Commit: r3117: use same osm id for way that is added with makeCycleWay().

Hi Gerd,

it is a bit better with --preserve-element-order but still not a complete solution. I have uploaded bigger sample from Paris: http://files.mkgmap.org.uk/download/186/29467175.osm.pbf

Please look at node: http://www.openstreetmap.org/node/2650495625

-- Best regards, Andrzej _______________________________________________ mkgmap-dev mailing list

mkgmap-dev@.org

...

http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

_______________________________________________ mkgmap-dev mailing list

mkgmap-dev@.org

http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

cycleway-v3.patch (4K) <http://gis.19327.n5.nabble.com/attachment/5800670/0/cycleway-v3.patch>;

-- View this message in context: http://gis.19327.n5.nabble.com/Commit-r3117-use-same-osm-id-for-way-that-is-... Sent from the Mkgmap Development mailing list archive at Nabble.com.

Hi Andrzej, did you try patch v3? I think it should work better than v2, because it makes sure that the WrongAngleFixer does the same for original way and cycleway Gerd popej wrote

Hi Gerd,

I think GPSMapEdit adds a node for each point that belongs to multiple roads. I have forgotten about this peculiarity.

You are right, img without cycleways contains the same points as img made with patch v2. I assume patch is correct.

-- Best regards, Andrzej _______________________________________________ mkgmap-dev mailing list

mkgmap-dev@.org

http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

-- View this message in context: http://gis.19327.n5.nabble.com/Commit-r3117-use-same-osm-id-for-way-that-is-... Sent from the Mkgmap Development mailing list archive at Nabble.com.

Hi Andrzej, popej wrote

Side note: simplification, that you are using, is quite light. It is good, but sometimes I would prefer more simplification to get lower size of a map. An option would be handy. Or maybe reduce-point-density and reduce-point-density-polygon are already active for resolution 24?

well, I called the branch high-prec-coords, so my goal was to get closer to the precision in OSM. The simplification is only a side effect, and for small polygons like single buildings it has no effect. On the other hand I see no reason to disable simplification for resolution 24 as we have it now, we just have to make sure that we don't remove info that is needed for routing . I add this to my todo list. Gerd -- View this message in context: http://gis.19327.n5.nabble.com/Commit-r3117-use-same-osm-id-for-way-that-is-... Sent from the Mkgmap Development mailing list archive at Nabble.com.

Hi Andrzej,

I have tried v3 and I haven't found any problems yet. Shape of roads is correct, opposite cycleways are created, long distance routing works in BaseCamp.

fine :-)

I wasn't able to create bicycle route on opposite cycleway in BaseCamp but it works in Mapsource.

I'll try that tomorrow.

I have noticed one unexpected effect. Tiles in England are marked as ZIP after city while tiles in France get ZIP before city flag. I can't guess what is controlling this flag. Is it style dependent or hard-coded? Anyway, would be grate if you could add the same support for dirve-on-left flag.

The configuration file is mkgmap\resources\LocatorConfig.xml , search for streetBeforeHousenumber and postalcodeBeforeCity. I think it is okay to use that also for drive-on-left flag. This was discussed before, the problem is that one tile can contain different settings, and we do not (yet) support polygons for boundaries. I've also noticed a discussion about the tag driving_side, see http://gis.19327.n5.nabble.com/Driving-side-tp5800583.html Gerd

Hi Andrzej,

...

the problem is that one tile can contain different settings

Isn't it exactly the same for ZIP flag?

yes, but I think a routing error is more important.

Actually most tiles don't contain different settings. I'm creating map of Europe where a few tiles for over thousand have mixed settings. So definitely preconfigured setting for country would be better then single default for all.

Maybe I'm wrong, but it seems so easy to add dirve-on-left flag to LocatorConfig, that I can't stop myself from pressing for this solution ;)

yes, I agree that this is the better solution, but I am not familiar with that part of the code. If I got that right, the flag is only important for roundabouts, and we already discussed that we should count them and - if a tile contains different drive-on-left values - use the one with the higher number for the complete tile. I'll try to code it like that in the next days. Gerd

Hi Andrzej, GerdP wrote

...

I have tried v3 and I haven't found any problems yet. Shape of roads is correct, opposite cycleways are created, long distance routing works in BaseCamp.

I think I found a cleaner solution. I'll call makeCycleWay() from StyledConverter.convertWay(). That removes the need to save the copy of the way with a different id. I have to make sure that this doesn't produce side effects. If not, I'll post another patch tomorrow. Gerd -- View this message in context: http://gis.19327.n5.nabble.com/Commit-r3117-use-same-osm-id-for-way-that-is-... Sent from the Mkgmap Development mailing list archive at Nabble.com.