Yes, preprocess if it's not done in gosmore. Inside Gosmore it would be "on the fly".
On Fri, Sep 12, 2008 at 12:44 PM, Lambertus <[EMAIL PROTECTED]> wrote: > This sounds like a lot of work, adding all those extra tags to the (already > existing) data in the OSM database. Or do you propose to preprocess the > planet data to add those tags with some script just before a Gosmore > database rebuild? > > Simplifying this would be to preprocess the planet file to add maxspeed=1 > (or something like that) to any road that is not in a cycle relation before > sending it to Gosmore. But this and my previous solution work for YOURS only > because it doesn't need the Gosmore UI and can easilly use multiple > instances to provide task specific routing. Your solution could be a good > long-term solution where both the Gosmore routing and UI would properly > handle this specific routing scenario. > > > Nic Roets wrote: > >> The best solution is to >> 1. create new way types e.g. in addition to 'highway=residential' etc, to >> also have cyclenet=residential etc. >> 2. Create entries for these in elemstyles. The rendering variables would >> be identical, but they will differ substantially in the <route >> bicycle="..."> field. >> 3. Write software to map highway keys to cyclenet keys where appropriate, >> e.g. when rcn is set or the way is part of an appropriate relation. I have >> ideas how to do this in gosmore, but haven't found the inspiration yet. >> >> This way gosmore will route over non cyclenet ways but will prefer >> cyclenets. >> >> Regards, >> Nic >> >> On Fri, Sep 12, 2008 at 11:47 AM, Lambertus <[EMAIL PROTECTED] <mailto: >> [EMAIL PROTECTED]>> wrote: >> >> After some more though, there is a 'solution' for this. I'm not >> sure it >> will produce the best results in every case, but it's worth a try. >> Consider this: >> >> House -> cycleways -> cycle routes -> cycleways -> Work >> >> The cycleroute instance of Gosmore given the House and Work >> coordinates >> will produce a route using cycle routes from coordinates nearest >> to the >> House and Work coordinates. We'll name these to and from coordinates >> House_cn and Work_cn hereafter. >> >> Then the route is calculated from House to House_cn and Work_cn to >> Work >> using the normal Gosmore instance. These three routes combined >> will give >> you a route from House to Work using cycleroutes as much and fast >> (closest) to your House and Work as possible. The only adjustment to >> Gosmore would be to persuade it to find the nearest way from a given >> coordinate even for large distances instead of the few dozen meters it >> allows now. >> >> Would this suit your expectations? >> >> Shaun McDonald wrote: >> > Lambertus wrote: >> >> Please note that the routing database is *only* filled with cycle >> >> routes that are mapped using *relations*. So there is no routing >> >> using cycle routes rendered on the Cycle map layer mapped with tags >> >> like 'ncn_ref' only... >> >> >> >> Unfortunately you can't see the difference on the cycle map, >> but when >> >> the website reports: 'Unable to calculate a route' then you can be >> >> pretty sure this is because of the above. >> >> >> > >> > This explains why I can't route from my house to work, where at each >> > end there is no local cycle route, and there are several small >> bits in >> > the middle where you cannot route over. This is either due to me not >> > having bothered to split the way, or as you say, there is a break in >> > the relation. >> > >> > It would be really cool if you could have an edit permalink so that >> > you can edit the data on the openstreetmap web site. >> > >> > Would it be possible for you to have a different marker for the >> start >> > and end? >> > >> > Finally I was initially expecting a routing system that would prefer >> > cycle routes, however this makes an excellent data >> validation/testing >> > tool. >> > >> > Shaun >> >> >> _______________________________________________ >> talk mailing list >> [email protected] <mailto:[email protected]> >> http://lists.openstreetmap.org/listinfo/talk >> >> >> >
_______________________________________________ talk mailing list [email protected] http://lists.openstreetmap.org/listinfo/talk
