|
On 2014-03-29 14:13, SomeoneElse wrote
:
On 29/03/2014 12:41, nounours77 wrote:Unfortunately, this is the kind of fuzziness that makes GPSes send cars to forbidden places or through mud (1) or hikers on a 5 km useless detour, that makes people laugh at OSM users, that makes OSM taggers laugh at themselves and laugh at me when I say that routing is a prominent application of OSM. That disparages OSM as a whole. Different features have different degrees of importance. Mapping every details like trees and their species is adorning and less important. But mapping the features that tourists look for like Nounours wants to do or road hazards, especially to spare a child's life while looking for the features, like I want to do are important, and both are disregarded. I have tried to show that renting is akin to selling, that they fit in the same framework, that if you have car renting defined and you want to support boat renting too, you almost just add the word "boat" to the framework (like reusing an object in object oriented programming) and that this lessens the fuss of voting new propositions. No one seems interested. I also had rendering problems. In the same reasoning vein, I suggested to use object oriented like generic rendering (e.g. landuse=leisure) that would be used if no particular rendering exists (leisure:miniature_golf=yes). Such frameworks tend to have everybody think in the same direction. No interest. If you look at (random cases) associatedStreet relation or addr:country=*, some discussions will say that you should not use it and other discussions will say that you should, but the wiki is mute about that or almost. The reasoning about addr:country can be found under is_in=* which is an older alternative but none of them points to each other and it's not said that addr:country is better that is_in=* because it shows that the name is a country. In conclusion, half of the taggers will do it one way and half the other way. And as the discussions say that one of the ways is not supported by all data consumers, half of the tagging won't work for that consumer. What about everybody doing the same thing so that the consumer did the job only once, whichever way it is? Yes, Nounours is right. If tagging is not precisely defined, the taggers will tag each their own way and data consumers will not understand it and it will have been tagged in vain. It is true that some cases are less strict than others but the problem is that many taggers have a tendency to make no difference and tag everything à la Picasso. Last but not least, I think we forgot to thank Nounours deeply for the work he does. Or tries to do. Cheers,
(1) I had found (with Osmand too) and corrected several similar GPS routing tagging mistakes (there are many) and I was wondering why the same mistakes repeat over and over again. Then I found that the same mistake existed in my country's national wiki instructions!!! I put that right, but I was told off by someone standing himself as a chief and I was commanded to put the error back to the wiki because 1) no one would tag it that way (too complicated (3 tags)), 2) everybody knows that signal XXX means what I wanted to have the tags mean 3) there had been no discussion. A little bit of thinking leads to these conclusions: 2a: such tagging is not a matter of people but of programs understanding it, which I had corrected it for; 2b: if one sees tags, they don't say which road sign they describe, so that not even a human can interpret it rightly; 3: if a car is sent to where it shouldn't go by the tagging, replacing it with the tags that send it to the right way needs little discussion beyond "fine, thanks"; 1: if nobody will do it that way and wiki instructions are to not do it that way and OSM is laughed at and even laughs at themselves, well, how should I say, there is a problem. |
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
