Following from TagInfo's link, if we adopt the most used practice as de facto, that would be capital=[lowest admin_level of respective regions], by a large margin. However, its most used value (capital=8) makes little sense (that represents the capital of an admin_level=8 area, which would be the capital of a city!; I think those mappers meant "this capital is a city" :P). The "yes" value is used almost as often as "4", but then there are "6" and "7" (representing capitals of areas that are not really equivalent to "states") which make the "yes" value less common by far (yet those could also be the same mistaken case of capital=8).
I think "capital=[number]" is fine as long as people understand (so we should note that on the wiki article) that relations should be properly set up to refer to capitals as admin_centre (except where the two are different cities). If that the purpose of this tag is mostly (if not only) to help renderers do their job with lower processing complexity, I think we should discourage applying multiple values (e.g. "2;4;6;7;8"). Note: and in Brazil, we still need to discuss with the local community if it makes any sense to tag statistical regions (admin_level=3/5/6/7) without administrative or political power as boundary=administrative (perhaps boundary=political or a new value would make more sense) before we recommend tagging their main cities with capital=[number]. On Sun, May 18, 2014 at 9:43 AM, John Packer <john.pack...@gmail.com> wrote: > Honest question: are there capitals for something besides countries and > states? > > If not, we could keep it simple: > * capital=yes for country capitals > * state_capital=yes for state capitals (already in use in some parts of > America). > > PS: Taginfo just released some new functionality, which allow us to compare > the different uses of the different keys "capital" around the world: > http://taginfo.openstreetmap.org/compare/state_capital/capital_city/capital_level/is_capital/capital > (note that as demonstrated before, around 90% of the current values of the > key capital=* seems to be incorrectly tagged). > I'm not saying we should use the others, just though it was interesting. > > > > 2014-05-16 7:07 GMT-03:00 Martin Koppenhoefer <dieterdre...@gmail.com>: >> >> >> 2014-05-16 12:00 GMT+02:00 Andreas Goss <andi...@t-online.de>: >> >>> For what apart from Rendering will this "importance" be usefull? >>> >>> And I still think it would be missleading and will lead to confusion. >>> Someone who wants to look up all state capitals might just check all nodes >>> for capital=4 (because it's the logical thing to do) not knowing that it can >>> only be done correctly by also looking at relations. >> >> >> >> >> I don't see this as a problem. Looking for all state capitals is not >> within the scope of the capital tag, as you will miss those which are also >> capitals for bigger entities like countries. You might have to look up all >> capital=1/2/3/4/yes places and remove those which aren't state (level 4) >> capitals. Yes, rendering is probably the most important use case for this >> tag, why should that be a problem? >> >> cheers, >> Martin >> >> _______________________________________________ >> Tagging mailing list >> Tagging@openstreetmap.org >> https://lists.openstreetmap.org/listinfo/tagging >> > > > _______________________________________________ > Tagging mailing list > Tagging@openstreetmap.org > https://lists.openstreetmap.org/listinfo/tagging > -- Fernando Trebien +55 (51) 9962-5409 "Nullius in verba." _______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
