https://www.openstreetmap.org/changeset/94784598#map=11/1.5314/103.5724 See This, do you all think it is suitable to put place=neighbourhood for each apartment/ condominium development? And also for place=suburb for some bigger tamans e.g. Bnadar Universiti, Iskandar Puteri. Bandar Inderapura Kulai also. Some say should due to ease of nominatim.
Nominatim uses a ranking system for address calculations:
country = 4
state = 8
town = 17
Village, hamlet, municipality, district = 18
Suburb, croft, subdivision, farm, locality = 20
neighbourhood, housing estate = 22
street = 26
House, building = 28
so if its tagged correctly, then it should be ok
Tricky question. Personally I’m no authority to seek for the answer, anyway, here’s my take.
Nominatim used in OSM.org isn’t perfect, but to be fair, if someone wants to accurately display Malaysian addresses, the actual way out is create a custom Nominatim query specifically for Malaysia only (ahem, ideally by something like openstreetmap.org**.my** initiative, that is, for example).
I’ll take some examples from the UK, the birthplace of OSM: Sample 1, Sample 2, Sample 3.
Also from angys’s changeset discussion, “I just noticed this issue, like in Hong Kong they also dont add this place for neighbourhood.” I happened to check around the area, pointed by angys - and checked this out.
Phew. I don’t (and tried not to) map any closed way residential landuse (for taman perumahan, including apartments) with a place=neighbourhood tag. Sometimes I have to map for the renderer - these tamans were put as a place=neighbourhood node because several residential areas are next to one another, and I don’t have the knowledge of their boundaries.
Elaboration for the former case - some tamans could probably consist less than ~10 individual houses. It doesn’t make any sense that these tamans mapped as a place=neighbourhood, rivalling adjacent areas around it e.g. disorganised cluster of houses forming distinct small kampungs, likely unmapped.
Some cases considered, if a place is better known by the newer (more modern?) name due to an established sprawl of taman perumahan than the actual smaller villages or hamlets, then I might be inclined to map such places with place=neighbourhood.
Back to the actual discussion: personally for me I see this as trying to force Nominatim doing Malaysian addresses - mapping for the renderer.
Bigger taman as a suburb? Hmmm… I’ve no definite answer for this. I can only see Bandar Cassia, Penang fits the description (that’s what I can only think during this midnight hours). Otherwise, there’s no need to incorrectly map places especially if in reality, these areas are not that “significant”.
A closed way tagged as neighborhood ensures that nowhere else gets that address since the boundary is already defined. This should be easily and pretty accurately drawn for apartment complexes, a teeny bit harder for tamans, and hard for everything else higher in the hierarchy.
A node tagged as neighborhood makes unlikely places get tagged with its address since we don’t know how the boundaries are determined by nominatim/osm/whoever/whomever.
A closed, named landuse residential drawn from wherever (google probably) is just nonsense - the polite word - since it’s anybody’s guess where they actually should be. The same goes to district boundaries.