ตราสัญลักษณ์ OpenStreetMap OpenStreetMap

Routable Commons

โพสต์โดย CatastropheAsh เมื่อ 26 กรกฎาคม 2009 ในภาษา English

I went out on the bike and did Carseldine, Fitzgibbon, and a bit of Geebung. I think I've finished everything in the newer development areas around there where the satellite imagery starts to flake out. It's starting to look pretty too.

There's a few nonames and a fair few parks and things that might be better to do with walking papers, but other than that the area's looking pretty complete road-wise. It's been fun. :)

The issue that's concerning me today is to do with pedestrian routing. I've traditionally placed parks and commons a small distance away from the road, mostly because it just seems to be the done thing.

Nodes aren't touching, the park is an entity unto itself.

I'm wondering now if they should be "touching" in cases like this, at least in one point at either side. At the moment, it would take a seriously advanced algorithm to plan a pedestrian route through a park like this (and I doubt such a thing exists.) It would make perfect sense to walk through the park to get to the train station for instance, whereas walking around the road would take significantly longer.

Nodes aren't touching, the park is an entity unto itself.

The down side is you'd need an advanced editor like JOSM to successfully unglue and modify the ways if the time ever came. I'm loathe to stick things together like this (and I've only been doing so with landuse designations and the like, because it makes more sense to do so.)

So essentially I'm wondering what your thoughts on this issue are. How should we tag areas like this for pedestrian routing?

ที่ตั้ง: Somerset, Carseldine, City of Brisbane, Queensland, 4034, Australia
Email icon Bluesky Icon Facebook Icon LinkedIn Icon Mastodon Icon Telegram Icon X Icon

การอภิปราย

ความคิดเห็นจาก Baloo Uriza เมื่อ 26 กรกฎาคม 2009 เมื่อเวลา 18:00 น.

Find a path through the park, map that instead. Put the park boundaries at their actual locations, not the road centerline. Example: osm.org/?lat=45.52229&lon=-122.83711&zoom=17&layers=0B00FTF

ความคิดเห็นจาก Komяpa เมื่อ 26 กรกฎาคม 2009 เมื่อเวลา 18:01 น.

My personal opinion - you should paint a footway connecting both things. "Park" shouldn't be used for routing, highway=footway/pedestrain should.

ความคิดเห็นจาก HannesHH เมื่อ 26 กรกฎาคม 2009 เมื่อเวลา 20:15 น.

I am very pro sharing the nodes, but as you can read here this is a hot issue: osm.org/user/HannesHH/diary/3393

ความคิดเห็นจาก marscot เมื่อ 26 กรกฎาคม 2009 เมื่อเวลา 22:29 น.

yeah map path in park and join that to the road.and yeah put the boundaries as near to their right locations as you can.

ความคิดเห็นจาก Waldo000000 เมื่อ 26 กรกฎาคม 2009 เมื่อเวลา 22:47 น.

What if there isn't a path through the park? To me, it would seem strange for a router *not* to suggest a route through a park if on foot, provided there's no access restrictions...

ความคิดเห็นจาก BlueMM เมื่อ 27 กรกฎาคม 2009 เมื่อเวลา 03:05 น.

I'm personally in the "mark boundaries at actual locations" camp. I believe some routing programs allow for the non-joined nodes. I think of it as data accuracy, just because roads (areas) are modelled as ways, doesn't mean other features have to loose accuracy. Also, let's say someone comes along and asks "what is the total area of local parks & reserves in my local council?" -> it will be wrong if we share nodes.

ความคิดเห็นจาก StefanB เมื่อ 27 กรกฎาคม 2009 เมื่อเวลา 06:33 น.

So can we now expect areas of unroutable regions in the park, tagged access=no, landuse=rose_garden so that routing algorithm can find optimal way trough a park without footpaths (and whole park tagged with access=foot). Oh, and imagine the exploding size of routing matrix :)

ความคิดเห็นจาก davidearl เมื่อ 27 กรกฎาคม 2009 เมื่อเวลา 11:13 น.

As people have said, there's differences of opinions. But I'm in the "keep them separate camp" and put the paths in, even if it is only a short way through an entrance - someone will connect it up sooner or later.

There's two reasons, I think, one a point of principle and the other practical: (a) the road way is marking a centre line, and the park boundary does not share that centre line - on a typical 8m wide road with 2m each side of verge and/or footway, the park boundary will be often be 6m or more from the line marked out by the road, and that's a sufficiently large distance to be worth taking into account; and (b) it's much harder to edit ways which are overlaid on each other.

ความคิดเห็นจาก davidearl เมื่อ 27 กรกฎาคม 2009 เมื่อเวลา 11:16 น.

เข้าสู่ระบบเพื่อแสดงความคิดเห็น