Support addr:place in the Address preset

[johanneshoehn]

Some villages don't have street names, but use the place name instead of a street name in addresses.
This is very common in my region. For instance the village Hohlweiler belongs to the city Scheinfeld and has no street names. Addresses are of the scheme: (name), Hohlweiler (housenumber), 91443 Scheinfeld.

The OSM wiki recommends using addr:place instead of addr:street in those cases. iD only supports addr:street. It would be nice to be able to select wether to input addr:place or addr:street and even better if there was some kind of automatic detection that the name matches up with the place name.

[stephan75]

That whenever this feature request should be implemented, please have a validation that addr:place can only be added to an object when addr:street is NOT used. Because according to the OSM wiki definition these two keys should not co-exist on an object!

[MKuranowski]

I guess it could be solved by adding a tickbox which would swap addr:street with addr:place.

[RicoElectrico]

I second this as very much needed (as many as 6M uses of addr:place worldwide). There are many confused newbies, here's one: http://www.openstreetmap.org/changeset/47780046 . They always struggle with it, and it's not even their fault.

@bhousel If you need help on figuring out how to do this properly UX-wise, I think you may ask out on the mailing list. I don't know about other countries, but in our case it would suffice to have "place with street names"/"place without street names" option, or a tickbox with "address without a street name" which as mentioned would change "street" to "place".

[bhousel]

I don't know about other countries, but in our case it would suffice to have "place with street names"/"place without street names" option, or a tickbox with "address without a street name" which as mentioned would change "street" to "place".

Yes, I am ok with sticking a radio button somewhere at the top of the address block in the situation where a country has multiple alternate address formats. This is extra complicated though because iD would need to:

• Somehow add support for alternate per-country address formats to the address.json field schema (fine)
• Let the user tick the radio button to change the address field between alternate formats (easy)
• Detect how the radio button should be ticked when parsing existing address data (harder)

[MKuranowski]

So, I think:
addr:place == * && addr:street != * --> Place without street names
addr:place != * && addr:street == * --> Place with street names
addr:place != * && addr:street != * (or simply no tags) --> Place with street names, beacause this kind of places are more common.

The only problem comes when addr:place == * && addr:street == * (taginfo shows ~1M combinations).
UPDATE: Most of this cases looks like addr:place is mistaken with addr:city.

[rmikke]

I don't really think there is a need for alternate per-country address formats. In this wiki article all the examples use the same addressing format for addr:place. Ok, all the examples are from three countries only, but really, how many addressing schemas can you imagine? ;)

Now, maybe we should check if there are other cases for addr:place than villages? like e.g. some areas in the cities that have just common place name and number in addition to addr:city?

As for editing existing addresses - I'd suggest the same as Natsuyasumi. And if both addr:place and addr:street are present (or both absent) the radio button should be ticked for address with streets.

[MKuranowski]

@rmikke Of course there is a need for different address formats - you can see how Czech or Japan address formats are different from Polish or German formats.

In the article you provided there is a big bolded text explaining that addr:place does not mean a part of city - addr:suburb should be used instead.

If you want to check addresses with addr:place != addr:city you can play around with this Overpass query.
With such examples in Poland it simply looked like addr:city was mistaken with post office.
On the other hand, I have found some addresses in Germany (like this) where there were no street names in cities, so it means that there are places where addr:city can be different than addr:place, but this does not require any more options - it falls for the Place without street names scenario.

[stefanb]

We have the same situation in Slovenia. Smaller villages do not have street system and in these places the addresses are name_of_place + number, or in OSM schema: addr:place + addr:housenumber.

Currently the address format is

iD/data/address-formats.json

Lines 33 to 39 in 960e896

	{
	"countryCodes": ["at", "ch", "de", "si", "pl"],
	"format": [
	["street", "housenumber"],
	["postcode", "city"]
	]
	},

but that would need to be changed to easily support setting addr:place instead of addr: street.

It would be trivial to show both input fields, but users would be confused, and likely set them both to the same value, instead of leaving one empty.

A good compromise would be to introduce a new input type, eg streetOrPlace which would list nearby streets and places in the drop-down and autocomplete options, then set the appropriate tag automatically.