tOn 26/01/2016, Mateusz Konieczny <matkoni...@gmail.com> wrote: > In my experience name, name:en, old_name, alt_name, alt_name:ru etc etc > etc were always sufficient. An example where multivalue names are > truly necessary would be interesting.
Andy has already given some good answers and I've rambled for too long on the subject, but since you ask again I'll dig up http://www.openstreetmap.org/relation/5257865 again, which cannot be satisfyingly tagged with foo_name variations. Its name_1 and name_2 tags have absolutely no semantic difference, so puting them in foo_name and bar_name would be wrong. In fact even its name tag is semantically at the same level as the other two. The local knowledge comes from my in-laws, who lived there for as far as they can trace back. You might want to brush this example off as too rare to bother about, but I've stumbled uppon places with many names before. I used to agonize over the decision of what to put in which foo_name tag (alt and loc being the most likely candidates), with the result that I was assigning semantic value ("this name is only used locally, this one is a bit broader") when there was actually none. I'm sure I'm not the only one in this situation. We've trained mappers to always prefer foo_name but this is often wrong. We need multi-valued keys to accurately describe the world. Until the OSM data model supports that natively, we make do with either the suffix or the semicolon hack (plus some niche and one-off solutions). They are not great but they are necessary, embrace them. And while you're at it, recognize the cases where semicolons are problematic and embrace suffied tags. _______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
