Am Mo., 8. Okt. 2018 um 11:49 Uhr schrieb Dave Swarthout < daveswarth...@gmail.com>:
> So I retagged them this time assigning the name "Three Lakes" to the > relation containing the lakes and removing it from the individual lakes. I > checked back again today, after a few days had elapsed, and was able to > seach for and find Three Lakes. But also today, when I was checking the > area in order to use it as an example for my reponse to this thread, I > noticed that two of those lakes have individual names. I added the names > but JOSM complained. Of course, it wanted something more than just a name. > I added water=natural to each of those two lakes and was able to upload > them successfully. So now my lakes are tagged with natural=water twice, > once on each of their outer ways and once in the relation. If the > natural=water tag is removed from the relation, the simple lake doesn't > render as water. Where is the proper place for the natural=water tag? > IMHO the natural=water should at the very minimum go on the individual ways representing the single lakes (or relations if the lakes are mapped as multipolygons). For the group of lakes mapped as multipolygons, it seems natural that they do not inherit any water-properties any more, so if the name is referring to a group of lakes, this should be expressed explicitly, NOT by having simply lakes in a relation with type=multipolygon and name=* as the only tags (this would mean you don't know what this represents). An idea could be "natural=group_of_lakes" or "waterbodies", or "series_of_waterbodies", etc. Or you could add natural=water another time (it is not impossible to evaluate, but it is not what people expect to find I guess). The fact that natural=water does not render on a way if the way is part of a multipolygon relation (if I understand you correctly here: "If the natural=water tag is removed from the relation, the simple lake doesn't render as water.") seems to be an error (in osm2pgsql or the rendering style). but I did not observe this so far. Also, if JOSM complains about a mapping style you consider valid, you should raise an issue (it is not completely rare, the Josm validator is sometimes a bit overeager and has a tendency to nanny its users). > > I fool around with the tagging every time I create one of these monsters. > Just last week I created another multipolygon relation to hold a pair of > calderas named "Twin Calderas", one of which contains a lake. I searched > for it today and found it but another mapper had edited it in order to "fix > old style multipolygon". I asked him in a changeset comment what exactly he > had done to fix it and he replied, "The old-style multipolygon (with only > type=multipolygon tag) with the two members on the left (one outer and one > inner) was removed, the correctly tagged multipolygon was updated to also > include the inner from the other multipolygon, because it also includes its > outer." I'm reluctant to ask him to explain further because I'm clearly > confused by his answer and indeed, by the entire concept. > basically, the old style MPs had a magic way of interpreting both, the tags on the ways and on the relations, in a combined fashion. The new style MPs go for a cleaner and more transparent approach: tags apply to the objects to which they are attached. A tag on an outer way is for this way, a tag on a relation is for the relation (e.g. the area(s) with maybe holes, that a multipoligon relation represents). Cheers, Martin
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
