In tens of places in the Israeli map, the names of a places appear twice, one for the place node, and another for the residential area.
For me it’s a big eyesore. I don’t see it done in European countries much.
Should we remove the name of the residential area? Maybe move it to the note tag?
The node should have the “name” and “name:en” tags, and all the name* and name:* tags.
The polygon should only have the “place_name” and “place_name:en” tags.
I have done that in the last month when I added boundaries for each village / city.
I wasnt sure because of the “name” “place_name” tag, so I added both. I think this could be something we should repair by using a script.
Also because I never added all the alternative names from the place node to the boundary.
BTW: Because I couldnt be sure of the places I also kept the place node.
Don’t delete the nodes.
They are essential.
They tell the renderer where to draw the name of the place.
They also have all the tags related to the place, like population, wikipedia, etc.
The polygon should only have:
landuse=residential
place=*
place_name=
place_name:*=
All the rest (including the “place” tag) belong to the node.
It’s a manual job, as there are many permutations.
We have about 300 such place polygons, and I’m more than half way through.
Hope to finish by tomorrow.
There haven’t been many place nodes deletions.
The biggest I remember is a new mapper deleted Yeruham, but it was reverted in a short time…
Some villages here and there were deleted as well, but nothing too big that couldn’t be reverted.
Actually, I’m happy to say that we haven’t had many deletions in general, and those that had a reason behind them (and not by mistake) - Those reasons were discussed in this forum.
I think that there should only be a single entity tagged with “place=”, otherwise all such entities should belong to a relation. For example, this creates problems with address-finding services. Once they discover a place=* node not connected to a relation, they do an automatic interpolation for the extent of this place, even though there exists a boundary polygon somewhere. This results in weird things like one town being inside a larger city etc.
I did this for Ness-Ziona and Rehovot, and it seems to have solved the problem. I used these guidelines.
I stated my opinion above.
I formed my opinion based on this discussion.
It hasn’t changed in years, and I believe this is the current standard.
Actually, as long as we have a standard, it would be easy to change things with an automated script.
The current “problem” is that tagging isn’t standard.
I’m trying to standardize things, and once we have a clear and definite decision - it would be quite easy to change.
I need a small clarification on the current standard. How to tag polygons with landuse=* or boundary=* belonging to a place, in the case there are several of them? In particular, should all of them (which don’t have a name of their own, like a suburb or a named commercial area) have place=, place_name=, place_name:en=*?
landuse=residential should mark the built area.
boundary=* should mark the (political) administrative boundary.
They are not always the same.
suburbs should preferably have everything: A node and a polygon.
commercial areas - Probably only a polygon.
It doesn’t stop there! What about schools, hospitals, universities, forests, national parks, playgrounds?
I guess that for all of these, a polygon should suffice.
If there is no name, then there are no name* tags on the nodes, nor place_name* tags on the polygons.
If a place has a single residential polygon, this polygon should have place=* and place_name*. Otherwise, any unnamed (not belonging to a suburb) landuse=residential should only have is_in=*.
In the “Otherwise”, do you refer to the node? polygon?
The “is_in” is problematic, as it’s defined as “free text”
Unless we define it to only have the name of the place that it belongs to? (and then we’d need “is_in:en” with the name:en of the place it belong to…)
I’d prefer we try not to invent new tags (although we might have to in this case)
In a utopic world we’d have something like “belnogs_to_id” with the id of the node or polygon this entity belongs to…