Can we legitimately take shortcuts when entering addresses to exclude common enclosing data?
Example: the dozens of houses in a tract all have the same city + state + zip; can I add just housenumber + street?
Do I have to include the city and state for addresses that are clearly enclosed in the given city?
This just feels like a lot of typing and extra data in the database, and I don’t know that it actually provides anything useful.
I believe the minimum is street name and house number. Most software can figure out the city based on the surrounding administrative boundary or distance from the named place. ZIP codes are not areas but lists so that is a bit harder for software to figure out (and for me when surveying on the ground).
I collect house numbers with OsmPad and my editing work flow is to bring the OSM file generated by that into JOSM where I can do a select all then add the city. I don’t usually bother with ZIP code because I can’t see that on the ground while doing my survey. I also adjust the locations to match the houses (as opposed to points on the sidewalk in front of each house) based on either Bing or Mapbox imagery. Then, and only then, I download the area from OSM, review for duplicates, etc. and then upload back to OSM.
With that workflow adding the city and, if you are sure of it, the ZIP code is pretty simple for as many address points as you may have collected.
I haven’t really used ID and it has been years since I used the first version of Potlatch so I don’t know if that work flow is possible in other editors.
I would say: This depends on the area / country where you add address data.
At least in Germany we have complete boundary relations for administrative and postalcode purposes … we really have postalcodes organized like boundaries around smaller places or some more inside of bigger cities.
So you can ask yourself: are the administrative boundaries complete for your area in OSM? If not, or only partially, I recommend to add add:city!
And I assume that postalcodes are different in real world in the USA than in Europe … like n76 said. So I would also add them to get better clarification in countries where postalcodes are not exacly like area boundaries.
AND: in Germany we had a really useful purpose for addr:postcode on each object: we were able to filter any object inside a unique postalcode area with postcode=XXYYZZ where addr:postcode was different than XXYYZZ in OSM data. Thus we found thousands of errors either on the single address objects, or even wrong geometry of boundary lines.
After finishing adding housenumbers with street names in an area, you can just select all added entries and set addr:city and addr:postcode for them all at once. At least that’s how it’s done in JOSM I don’t know how about iD or similar
Ok, so continuing on my trek for how to do this well. I’m using my local condo complex as my testbed for fooling around with this stuff, and I’m just not sure how to handle it.
This is the area (Salerno Condominiums, in Southern California USA):
Every set of buildings has ~10 apartments, each with a different house number, but i just don’t know the best way to handle this.
Tagging the building with the range of addresses [almost all of them]
- addr:housenumber 62-80
- addr:interpolation even
Adding individual Address items nearest that entrance to the building [upper center]
Adding a pile of Address items near a common central place in each building [right side]
The only one that looks good is the first, but it’s the one that encodes the least information.
The other two show little number all over, and it looks terrible but nevertheless enables the kind of searching for a street address that I assume the project wants.
How does one do this while getting past “don’t worry about the renderer” ?
A couple of points in response to various posts:
I agree that typically housenumber + street name is adequate for most addresses. I would only add additional information in cases of exceptions or ambiguity. Adding addr:city to every address will cause problems if the postal address boundary is different from the administrative boundary (very common in the UK), or if the boundary changes (not unknown), or is incorrectly mapped. In the UK it also makes sense to add postcode information, when available, as representing postcodes as polygons is fraught with problems, to individual address. In places were postcodes are less fine grain this may not be needed.
Apartment blocks, such as the condos mentioned are probably most easily handled using a single address node with interpolation. I think in my local area we try and put this on the main entrance for groups of apartments. I think I tried to document some of this on the relevant wiki page (http://wiki.openstreetmap.org/wiki/Tag:building%3Dapartments#Addressing). (If you have any other ideas what could be added here, let me know)
In my country (the Netherlands) the problem was solved with a large import of all adressess of all buildings. Being a small country with a high population/building density this was quiet a task that took about a year to complete. The data came from the official building registers that are being kept in our country.
In the editor (an apartment with 20+ numbers) it looks like this:
On the map you see this:
So in the end there is a node for every address!
Maybe it helps?