Thank you for your thoughts and for tidying up an area within my example.
Firstly, I wanted to clarify that yes, we are talking about the situation when the golf course and residential area are all part of the same organization, and owning property does confer access rights to the golf course, much like not owning property precludes such access (unless one is a guest of a member). We can call these golf communities if the term “country club” is too geographically specific and, perhaps, a bit too broad as I imagine there exist country clubs in the US or elsewhere that do not include a golf course.
So, speaking of golf communities - I would be extremely reluctant to create two overlapping polygons for the residential community and the golf course. Primarily, that would defeat the purpose of defining an object as “golf community”. It would define a “residential community” and a “golf course”, not a “golf course with a residential community built into it” (or vice versa). Secondly, it could lead to rendering issues if we were to tag the name on both the residential polygon and the golf_course polygon. Not to mention it’s too much work ant twice the number of nodes not that anyone is counting.
Note that in the example from the wiki discussion you linked, the “good (and early) example” of tagging such community in England does not use an additional polygon for the residential, it simply calls the whole thing a golf_course. Granted, they have considerably fewer houses on the golf course than in my examples from Southern California.
I wanted to be clear on what you meant by
What’s an operator tag and which one did I suggest adding? I am relatively new to OSM and some of the lingo tends to stump me.
Having considered your suggestion, my own experience, the brief discussion on wiki and other people’s attempts, I would say this:
- Draw a polygon around the entire golf community and tag it landuse=residential; leisure=golf_course; access=private; name=XYZ Country Club.
- (Optionally and time permitting) Create a multipolygon relation and cut out as many “inner” holes as there are distinct and sizable pockets on the aerial image showing homes and roads as opposed to fairways, roughs, hazards, greens and such.
- Draw an outline of as much of the golf course as you can with one polygon, tagging it leisure=golf_course. What about the rest of the golf community you are leaving outside the polygon?
- Draw multiple polygons where you see the parts of the golf course on the aerial image tagging them all as leisure=golf_course; name=XYZ Country Club Golf Course. Rendering looks terrible with the name of the club repeated over and over on each green area.
Also, when you refer to my last method (using multipolygons and adding holes for the non-golfing pockets in a golf community) you talk about me being at the mercy of Mapnik. Are you suggesting that all multipolygons may break when pulled into a conventional GIS tool? I tend to treat multipolygons as an established and useful OSM tool. Any renderer, official or not, that does not respect them, would be severely crippled. But even so, following my DOs would result in the best possible representation even by a compromised renderer.
The only really crucial rendering assumption I am making is that a golf_course tag trumps the residential tag in shading.