Personally, I think this is still a sort of kludge, although no worse than the ones I discussed in my blog pos <http://sk53-osm.blogspot.com/2020/06/housing-terraces-in-wales-minor.html> t.
I'm aware of a number of terraces which are discontinuous, demonstrating that individual houses in a terrace are not building:part. A typical example would be a terrace bombed in the war where the bombed out houses were not replaced. There is a "terrace" in Richings Park, Iver which looks just as if such a scenario had occurred, however, the owner of the end house explained that the developer ran out of money and never completed the terrace (the end houses were planned to be fancier). At one stage I terraced buildings and left the outline of the terrace as well as the individual houses which was a similar solution, but that will now lead to lots of error messages. For S3DB (simple 3D buildings) describing the entire terrace in terms of roof shape etc is often far easier than doing it for individual houses, so there are other advantages. My main objection is that it is not semantically accurate. The use of building=terrace both for entire terraces and individual houses in the terrace also is something I would like to disambiguate. For instance use building=terrace for the entire terrace & building=terraced_house for individual houses in a terrace (this latter value may also work with building:part in ways that give data consumers flexibility with the data). Generic building=house is preferably avoided for something more precise (detached, semidetached_house etc). I'd like to mark bungalows separately as, at least in Britain, they tend to be a very distinct housing type which building:levels=1 does not guarantee, but in various places, notably Southend, there are masses of semidetached bungalows. On the topic of the OP, I'm broadly with Chris on this, pretty much as I set out <http://sk53-osm.blogspot.com/2013/08/pfafing-about-opening-uk-address-data.html>7 years ago! I also think it's important that, for me at least, we're not adding addresses in OSM just to create an open replica of PAF. There are numerous other important uses of addresses over and above this and routing. At the Open Addresses meeting <http://sk53-osm.blogspot.com/2014/09/openstreetmap-at-uk-open-addresses.html> back in 2014 I participated in a discussion on this very point, and a number of people from large well-known organisations provided a good number of significant examples. I can't be more explicit because the meeting was held under Chatham House rules. If we do need to add postal towns, which I suspect we don't, then I would advocate for a specific tag addr:postal_town or even addr:rm_postal_town. In practice I would think postal towns can be deduced from post codes (i.e. externally to OSM): wikipedia certainly have lists for many postcode areas. Jerry On Sun, 20 Dec 2020 at 19:23, ndrw <nd...@redhazel.co.uk> wrote: > On 20/12/2020 18:44, ipswichmapper--- via Talk-GB wrote: > > What you do is give the outline way "buildong=terrace" and > > "name=<NAME_OF_TERRACE>" and all the houses with > > "building:part=house". The software can then tell that all those > > houses are part of the terrace called <NAME_OF_TERRACE> > > This is a good solution. I usually resort to simply not terracing the > building and adding addresses as points and/or an addr:interpolation line. > > In either case, if the name of the building is a part of the address > ("dependent thoroughfare") there is currently no suitable OSM tag for > it. I've seen cases of addr:place, addr:substreet or addr:parentstreet > but there is no established consensus yet. > > ndrw6 > > > > _______________________________________________ > Talk-GB mailing list > Talk-GB@openstreetmap.org > https://lists.openstreetmap.org/listinfo/talk-gb >
_______________________________________________ Talk-GB mailing list Talk-GB@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-gb