Hoi, I like to suggest the following structured approach to tackle the multi-value (MV) topic:
Phase 1: Are MV necessary? First, collect examples of seemingly necessary MVs in a wiki page. Classify them as ordered and unordered MVs. Discuss alternative tagging possibilities, which don't need MVs, for these examples. Try to figure out if MVs are rare or common. Discuss the variants of name=* separatly because they are most controversial. After the situation is clearly laid out, the community should agree if (ordered, unordered, both, no) MVs are necessary. If they are considered unnecessary, goto Phase 4. If they are considered necessary, goto Phase 2. Phase 2: Implement MV in the key- or in the value-domain? MVs can be implemented in the key-domain (e.g. _1 suffix) or in the value-domain (semicolons), or multiple identical keys could be allowed (only for unordered MVs). Discuss these three approaches, without discussing concrete implementations (e.g. whether _1 or %1 or whatever should be used). This discussion will also have to balance between a user burden (value-domain) and a programmer burden (key-domain). This might depend on the assumption if MVs are a rare thing or if they are common. Without wanting to stress the already questioned voting procedure even more, but normally, one would want to ask for agreement of the community here as well. If multiple identical keys should be allowed, goto Phase 4. If MV should be implemented in the value-domain, goto Phase 3. If MV should be implemented in the key-domain, goto Phase 3. Phase 3: Which separator to use? Key-domain: Decide how to separate the key's name and the unique suffix. Furthermore one should discuss what suffixes should be used. Value-domain: The semicolon seems to be the accepted separator in this case, but the whitespace after the separator might not. A relevant aspect in this discussion seems to be if these suffixes or separators are supposed to be presented to the user or hidden and automatically handled by the interface software. Goto Phase 4. Phase 4: Transition plan At this point, all decisions were made and the community agrees if and how MVs should be handled. Now, the transition from the current situation to the desired situation needs to be discussed. Depending on the outcome of the above phases, the transition could include changing editor software, but as well it could include re-tagging current MVs in a non-MV way. In any way, the time frame for such changes will be long. I suggest, that these phases are processed one after each other, because otherwise we will hardly agree on the ``higher levels''. Thus, now we should start to focus on phase 1 only. The different opinions that will likely show up there, will explain most of the disagreement that makes our current discussions, which mix all in one, so difficult, I am sure. It is already valuable if we could agree in phase 1 although we might not agree in phase 2. Mixing the discussion all in one prevents us from having this success in such a situation. I am willing to set up and care for a wiki page for phase 1, as such a page seems not to exist yet. (<http://wiki.openstreetmap.org/wiki/Multiple_values> documents the current state, rather than being a tool to support discussing the future.) As I'm relatively new to OSM (started mapping about a year ago, although my account is older), I don't want to force my way of thinking onto the community. In my opinion the approach I proposed could help, but if the OSM community is successful by doing it differently, I rather stay quiet and watch things evolve. Hope to help. meillo _______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
