[Osmf-talk] Question to board members

[Kai Krueger]

On 8/24/2012 9:25 AM, Mikel Maron wrote:
> >All I'm saying is that we must not focus solely on the map - it's bad
> > enough if our mappers already do that ;)
> Ok, let's focus on routing. There's been discussion and agreement 
> before (several years now)
> that having routing integrated in OSM.org in some way would be 
> valuable to mappers
> because it allows mappers to check the viability of OSM data for routing.
> Tiles and nominatim integration on OSM.org provide a similar function.
>
> Routing has not been integrated ad hoc. It's complicated both in terms 
> of back end and interface.
> We need a realistic and practical plan for routing to be integrated 
> into OSM. We need to try another strategy.

In recent times (and before that) Dennis Luxen (and others) has done 
some great work on Open Source Routing Machine and has turned it into a 
viable backend solution. With http://map.project-osrm.org/ he has even 
demonstrated the feasibility of operating a world wide daily updated map 
(at least for car routing). With that, the specs for the necessary 
hardware should also be now known. OSRM appears to support a pretty good 
selection of routing related osm-tagging and produces high quality 
routes in most cases if the data is correct. It would be therefore a 
good solution for routing on osm.org

If I understood it correctly he is also working on making it feasible to 
have multiple modes of transport, such as bike and pedestrian routing 
(without excessive hardware requirements). In the meantime one can 
easily use a backend like gosmore to supply those kind of routes. While 
gosmore can't do very long distance routes like Los Angeles to New York 
or to some degree even struggles with routes like Munich to Hamburg and 
thus isn't great for car routing, it handles short distances which is 
the norm in pedestrian and bicycle routing just fine with relatively 
moderate hardware resources. As an alternative, or in addition, one 
could probably also use an external routing backend like MapQuest for 
bicycle and pedestrian routing until osmf can take over operating that 
service itself.

With http://routing.apis.dev.openstreetmap.org/ there is also a 
functional prototype for integrating routing into the osm.org. It 
currently uses map.project-osrm.org as the default car routing backend 
and gosmore as the default backend for other modes of transport. One can 
also select MapQuest and CloudMade as routing engines to compare them.

Therefore I think that all of the necessary components to integrate 
routing into osm.org are now available. So as soon as OSMF buys the 
necessary hardware, one could probably integrate routing into osm.org 
comparatively quickly and easily.

Kai



>
> The most promising strategy right now is the model we employed for the 
> redaction bot, hiring someone to lead the effort.
>
>
> * Mikel Maron * +14152835207 @mikel s:mikelmaron
>
>     ------------------------------------------------------------------------
>     *From:* Frederik Ramm <frederik at remote.org>
>     *To:* osmf-talk at openstreetmap.org
>     *Sent:* Thursday, August 23, 2012 3:10 PM
>     *Subject:* Re: [Osmf-talk] Question to board members
>
>     Hi,
>
>     On Thu, 23 Aug 2012 22:13:18 +0300
>     Rich <richlv at nakts.net <mailto:richlv at nakts.net>> wrote:
>     > we don't send a message like that, it's just human nature - no
>     > reward, no motivation.
>
>     Yep. I'm all for more rewards! But those can come in many forms, and
>     Jaakko was specifically saying that people don't map what is not SHOWN
>     ON THE MAP, which is only one of many potential forms of reward.
>
>     > in a similar manner i have every now and then not mapped things i
>     > know won't show up.
>
>     That's because the map is a nicer reward than, say, some
>     simple statistic saying "Rich has contributed 1024 nodes today". But
>     there could be many other forms of reward, for example nicer and more
>     meaningful statistics broken down by areas or whatever. "Rich has
>     contributed 25% of routing-specific data in the area you are
>     viewing". (Where routing-specific data = turn restrictions, speed
>     limits, access restrictions - all hugely important stuff that is
>     difficult to show on any map, even at z20.)
>
>     All I'm saying is that we must not focus solely on the map - it's bad
>     enough if our mappers already do that ;)
>
>     Bye
>     Frederik
>
>
>     _______________________________________________
>     osmf-talk mailing list
>     osmf-talk at openstreetmap.org <mailto:osmf-talk at openstreetmap.org>
>     http://lists.openstreetmap.org/listinfo/osmf-talk
>
>
>
>
> _______________________________________________
> osmf-talk mailing list
> osmf-talk at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/osmf-talk

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/osmf-talk/attachments/20120824/22c909e1/attachment.html>