Gervase Markham wrote: > Which should I have done? That's the question I'm saying that anyone who > wants to extend a formerly binary tag with new values needs to provide > an answer to before they start using the new values.
IMO, one should implement the even-more-defensive option C, which is: if "bicycle=no": don't do it else if "bicycle=yes": do it else: do whatever makes sense for your map here It is an incorrect assumption that the "binary" tag only contains "yes" or "no" (or a handful of variants) when an editor is allowed to enter arbitrary values. And assuming there are a fixed number of values for "bicycle" is (in some ways) just as bad as assuming there are a fixed number of values for, say, "highway". ("Well, it's none of these other things--it must be a motorway!") > My point is that (leaving aside the specific bicycle example) extending > already-used tags in this way is going to result in confused renderers > and undefined and renderer-specific behaviour. I'd say this is totally true, in theory. But since we can't possibly stop people from extending tags (all they have to do is use Potlatch) we have to make sure the renderers can handle it. If someone edits the map to say "bicycle=masochistic" and a renderer draws it as a friendly bike route, it's the renderer's issue, ultimately. -Beej _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk