I know it remembers it. This is not my point. I dont want this for me.
I believe it would be better for new users to get a foot routing or get asked which mode they want to route.
Also foot routing fits better to the first impression of osm.org as the OSM Carto style is not car specific but mode neutral. Everybody walks. Not everybody drives cars. Foot is the most neutral mode of transport.
The reason it defaults to a fairly typical car profile is the Principle of Least Surprise. Most directions implementations, absent personal information about the user’s preferences, default to a car profile. I didn’t see a compelling reason to vary from that even though my personal preference is that every single motor car on Earth should be crushed into a small cube and launched into orbit around a far-distant planet.
You’re welcome to take up a whole bunch of discussion time by opening an issue somewhere, but I’d generally suggest that the way to improve the cause of pedestrian routing would be by finding a routing engine codebase and spending some time improving the profile code such that OSM becomes an even more compelling proposition for pedestrian routing.
Also foot routing fits better to the first impression of osm.org as the OSM Carto style is not car specific but mode neutral.
Strongly disagree. osm-carto is really very car-centric. It doesn’t show bike routes or bus routes at all. Roads are shown with much more prominence than paths or railways. Access restrictions for bikes and foot use are shown minimally or not at all. It doesn’t have the topographic information (hillshading/contours) that’s essential to anyone using their own two legs to get around. Car parks are shown before pubs. There’s a design rationale behind all of this and that’s fine, but I wouldn’t at all say that osm-carto is mode-neutral.
I was curious how it got to the current default router and you answered that for me. My ambitions to change this are not very big. I think i will leave it at that. I also don’t know the expectation of routing users, whether they are actually less surprised by car-routing than foot-routing.
Whether OSM Carto is now mode neutral or not, we disagree.
Yes, it shows car parking, but it also shows bike parking, bus stops and benches.
Yes, no bike and bus routes are shown, but no car routes either.
Yes, no bicycle or pedestrian access restrictions are shown, but none for cars either (motor_vehicle_no).
There is so much effort spent on PT routes, and use still way off.
I have to change a cycle route in my vicinity to more completely go around hurdles that recently got installed by the railway company.
Indeed, a previous change to that cycle route showed an obvious uptake in traffic of people with navi on handlebar. Fortunately, they do not have to rely on OSM Carto. @Nielkrokodil - And that is fine as is, isn’t it?
The website’s routing widget is more of a debugging tool than a showcase of routing and guidance (though it does help dispel the notion that OSM is just a pretty basemap). OSRM has its own demo map that exposes the router’s support for via waypoints and includes details like destinations and lane guidance (for the car profile). Most other routing engines have something similar. But the OSM website shoehorns every router’s instructions into a custom abstraction layer that only supports the lowest-common-denominator functionality common to every router, applying its own translations unrelated to the ones that the routing engines themselves use.
The principle-of-least-surprise profile for users in Vietnam would actually be a motorcycle profile – this being the country where motorroad=yes means the road is legally reserved for motorcyclists and maybe cars, per the blue Vienna Convention sign. Sadly, none of the major service providers offer one. Meanwhile, a pedestrian default would just sadden users in much of the U.S., where it would be too easy to end up with no route or an obviously dangerous one. The website’s routing widget could theoretically choose the profile most appropriate to the user’s country. But that would require IP geolocation or asking you and your browser for permission to access your location.
If we do want to showcase pedestrian routing, there might need to be a discussion about which routing engine provides the best experience for pedestrians. Despite my involvement in the OSRM project, I’m not certain it would be the best. If you’d like to impress people with OSM’s ability to facilitate pedestrian routing, there are some very good options outside of osm.org, such as Openrouteservice, which offers lots of options; OpenTripPlanner, which excels at multimodal routing (if you can find an instance that covers your region); a mobile public transit application such as Transit; or the informal transit apps by Trufi while we’re at it.