As previously mentioned in a previous conversation there are two elements defining Haifa’s boundary. (The specific polygons in from that conversation are gone). This is an issue that seems to come up often as in this conversation.
The residential zone seems to more accurately map the boundary of Haifa, both based on data from the city of Haifa and Wikipedia. I wanted to check before deleting the city way and moving its data to the residential relation.
They are both tagged wrongly. There should not be a massive landuse=residential area covering the whole city, as there are many areas within that are not residential. Usually a city would have a official boundary mapped as an admin area relation.
Last I checked, this is an issue affecting nearly all cities in Israel. I brought it up in the Telegram group back then, but didn’t get much attention and personally I don’t have what it takes to start fixing it. Fixing it will be a massive editing challenge, well worth the effort IMO but not a job for one person (not me at least). Needs some good coordination.
I guess mappers did this in the early days so the shape of the city shows up on a map.
Yeah, looking on overpass there are over 1200 named landuse=residential in Israel. I would say that where proper admin boundaries are defined for the city, town, the landuse=residential tag can be removed or deleted. Starting from scratch is the only way to encourage the creation of detailed, smaller landuses.Something like Jerusalem has duplicated boundary=administrative relations e.g. Relation: Jerusalem (1381350) | OpenStreetMap (real) and Relation: Jerusalem (6502363) | OpenStreetMap (fake).
When it comes to things like villages, I doubt the residential areas are official boundaries, so they should only be mapped with a place=village node, and then the residential area can be unnamed and split to exclude non-residential areas.
I was invested in this once and I don’t remember the nuances today. But I remember I was happy with how I fixed Daliat el Carmel / Isfiya and I suggested this as a model for a country-wide change
I recall looking over that conversation for context before starting this one. Though that was a problem here, now Haifa only shows up once Haifa | OpenStreetMap . What I was more interested in is the landuse as mentioned here : Places are rendered twice - #7 by wowik and what I have been implementing in and around Haifa
Although I generally agree with the idea of not surrounding a place with a massive landuse=residential, the boundaries of this polygon are now meaningless.
way/60356145 was originally mapped to be where the physical crawl of the buildings roughly ends. Without the landuse=residential this has no meaning anymore.
This issue isn’t haifa-specific. I think the current practice and the IL place Wiki guidelines are not optimal. I think we should agree on how exactly we want to deal with places. The tagging among the following elements needs to be very clear:
The giant landuse=residential which wraps many places. (My current thoughts: delete its place-related tags. Or delete it altogether and create more finegrained polygons)
The administrative boundary=administrative relation of a place.
The place point.
Where to put the place=* tag (Probably on the place point)
Dealing with cases that lack the boundary=administrative relation.
We should stick to worldwide best practices as much as possible. If I have the time maybe I’ll compare how different places do it in practice and suggest something concrete.
