streetcomplete
No housenumber task maps to wrong OSM tag.
See https://www.openstreetmap.org/changeset/73977621#map=18/51.51407/-2.54197
Not sure why noaddress is required -- National Rail seem to think it has an address?
it should have added "nohousenumber=yes"
From changeset comments it appears to be that the quest is "Does this building have a housenumber?" not "Does this building have an address?".
Adding "noaddress=yes" means the building doesn't have an address.
Either change the question or change the OSM tag that get's added.
Here's a (copyrighted) address for the station, i.e. it has an address, but no housenumber.
Hmm, makes sense. StreetComplete should use nohousenumber, even though this tag is barely used yet.
Note that the tag was just invented a few months ago. noaddress is in use for years already.
The change would mean, though, that buildings that not only have no housenumber but also have no address would also be tagged with nohousenumber because naturally, the surveyor has no implement to find out if a building has no address at all or "just" no house number if there is no housenumber plate.
Unfortunately, the use of nohousenumber=yes is often added wrongly by streetcomplete users. Example: https://www.openstreetmap.org/way/377703085 – This building almost certainly simply has no address. However, the tag nohousenumber should only be used for addresses which do exist, but omit the housenumber part:
For addresses which do not have a house number (for example "snc" = "senza numero civico" in Italian, "s/n" = "sin número" in Spanish and "sem número" in Portuguese) you can add
nohousenumber=yes.
(Note: I'm pretty sure this case just does not exist for addresses in Austria.)
Another related case would be where a building/address has an address, but it is just not signposted publicly.
So the issue here is that there are small hospital buildings which are the hospital itself and have a house number (and therefor should be asked by SC) or we have a big hospital complex where not every small building has its own address.
Perhaps we should drop building=hospital from the housenumber quest because we can not distinguish both cases.
Example: https://www.openstreetmap.org/way/377703085 – This building almost certainly simply has no address
Why are you so sure? Other hospital buildings inside that hospital complex do seem to have addresses (with housenumbers)?
So I wonder is that specific case an user error (i.e. that building does have an address with housenumber, but user answered wrongly), or StreetComplete lacking precision (i.e. that building does not have any address at all in reality [contrary to other hospital buildings in the complex] but the user only could tag that it has "no housenumber" or "leave a note", and chose the former as it was closest to "it has no address")?
(Note: I'm pretty sure this case just does not exist for addresses in Austria.)
Regarding nohousenumber=yes not existing in Austria (IIUC?), looking at taginfo, it seems that out of 2k values, about half of the values were two big imports (or otherwise mechanical edits) adding nohousenumber in Austria in last year or so?
Also, running overpass it looks like more of half of them are on building=yes or objects missing building tag, so those at least couldn't have been tagged by StreetComplete (IOW, if nohousenumber tag should not exist in Austria, the problem is significantly wider than just StreetComplete mistagging them)
or we have a big hospital complex where not every small building has its own address.
those should be skipped already:
https://github.com/streetcomplete/StreetComplete/blob/3dc1b77ed7d35c0907559207826e528945050e48/app/src/androidMain/kotlin/de/westnordost/streetcomplete/quests/address/AddHousenumber.kt#L88
But this building seems to be inside area which does not have address tagged: way 377703086
And that seems to be because other buildings inside that hospital complex do have their own separate housenumber (e.g. 45758601, 5625236, 45758602 etc.)
So, if anything, I would guess that this specific hospital building 377703085 should also have an regular address with housenumber (or if it does not, that it is an exception rather than a rule).
Perhaps we should drop building=hospital from the housenumber quest because we can not distinguish both cases.
Uh... maybe we should not? I'd really prefer if we recorded addresses for hospitals, it is really useful information. Even if rarely it might happen that some small hospital building gets mistagged with nohousenumber=yes instead of noaddress=yes.
What could be done to improve situation:
- split current Other answer
It has no house numberin "What's the house number of this building?" Quest to two:It has an address without house number==> tagsnohousenumber=yesandIt does not have an address at all=> tagsnoaddress=yes
and/or
- add the Other answer to "What street is this on?" Quest:
It does not have an address at all=> tagsnoaddress=yes(and removesnohousenumberif existing)
Somewhat related (using school as more common example):
- You have a school that is on one "Grundstück" (piece of land; property)
- Addresses in Germany are given to a Grundstück not a building, but usually you have only one building (residential, office, fire_station, ...) on one Grundstück (ignoring outbuildings).
- The school has 3 buildings; OSM: 2x building=school
Answer A:
- Some users answer with addr:housenumber for the building=school and nohousenumber for the other two buildings.
Answer B:
- Some users answer with addr:housenumber three times.
It is at least not clear to users what to answer in such cases.
(And the amenity=school area around the buildings can have an addr too).
(And the amenity=school area around the buildings can have an addr too).
In which case the quest is not asked.
It has an address without house number ==> tags nohousenumber=yes and It does not have an address at all => tags noaddress=yes
The user is asked about a house number. He is on-site and doesn't see a housenumber. How should he answer the question whether "it has no address" or "it has no house number", exactly?
However, the tag nohousenumber should only be used for addresses which do exist, but omit the housenumber part:
https://wiki.openstreetmap.org/wiki/Key:nohousenumber also has
For explicitly stating that no house number is assigned to a place. Show/edit corresponding data item.
are you sure that nohousenumber cannot be used for buildings which have no house number and also have no other address
(BTW, comments in old issues may be missed or forgotten - it is better to open a new issue if you are sure about the problem, then it needs to be deliberately closed for it to go away)
I opened https://wiki.openstreetmap.org/wiki/Talk:Key:nohousenumber#Is_it_fine_to_use_it_when_address_is_altogether_not_present%3F
Myself, I keep leaving notes "is there address for this building in official data" rather than selecting no house number answer
https://wiki.openstreetmap.org/wiki/Talk:Key:nohousenumber#Is_it_fine_to_use_it_when_address_is_altogether_not_present%3F has
And when checking the items entered by SC, there is a 100% error rate.
(which probably applies only in some areas, but still...)
I therefore believe that SC should never have suggested this tag, in neither of the two situations mentioned in the GitHub ticket, and should have simply added a tag describing the absence of the sign. Another source is needed to determine whether the absence of a sign means that there is no house number.
housenumber:signed=no sounds reasonable.
Back when this was changed, *:signed tag wasn't established at all. But I do recognize that the wiki already back then stated:
Do not use this tag if you are unsure about the housenumber or if there is a housenumber assigned but not signposted.
On the other hand, given that StreetComplete started using this tag to denote that there's no housenumber in September 2019, almost 6 years ago, it is clear that most probably a lot of the usage of this tag got added by StreetComplete
as the tag was basically unused before, hence, SC might have mucked this up and current usage of this tag might very well somewhat be defined by how SC uses this tag.
So, I am not sure whether it wouldn't be better to align the wiki with the de-facto usage of the tag instead, e.g. "Originally intended to exclusively denote that a house has no house number (but still an address), it's usage by StreetComplete since September 2019 effectively weakened its meaning to denote that there's simply no (visible) house number, without any further implications" or similar.
If so, I would use addr:housenumber:signed=no.
I would clearly prefer addr:housenumber:signed=no, as it fits in with the tagging in other quests (opening hours, max weight, name, etc.).
There's also noname for roads, though, when the user answers that a road has no road name (sign). 🤔 - should we reconsider there, too?
Maybe add option "this road definitely has no name" and "not sure, but no name is signed"?
(sometimes you can be 100% sure in villages and small towns, and noname=yes will be understood also by other tools)