[Talk-GB] importing house shapes

[Richard Mann]

That's pretty much entirely a relation-as-category, though, isn't it?

I'm wondering whether there'd be a case for a very small number of
high-value (in terms of processing speed) relations to be created
automatically, available to data consumers in the normal way through the
API, but _not_ shown in editors? These relations really just make things
more complicated.

Richard

On Fri, Oct 19, 2012 at 10:57 AM, Matt Williams <lists at milliams.com> wrote:

> On 19 October 2012 10:05, thomas van der veen <th.vanderveen at gmail.com>
> wrote:
> > I have used the terraced house plug in a few times and have found it very
> > useful, but I don't quite understand the difference between addr:street
> > (which is what I do at the moment) and adding a relation?
> >
> > Could you please explain this a bit more (and if possible in other
> contexts
> > then the postcode finder)
>
> Sure. All this addressing is based on the Karlsruhe Schema [1] which
> was devised as a way of tagging streets by some Germans who wanted to
> try out tagging a whole city. The original schema was a series of tags
> you put on each house such as house number, post code etc. The way
> they originally specified which street a house is on was by the
> addr:street tag where they would put the name of the street in that
> tag on each house.
> During the process of this schema evolving and being used, some people
> felt that this was redundant and so decided to start using relations
> to group these houses together (along with the street). I have a
> feeling this was in the early days of relations when many people
> weren't sure how to use them and so adoption of this method was slow.
>
> So, for adding in data as a mapper, it a difference between adding an
> 'addr:street' tag to every house in the street or selecting all the
> houses and roads in the street and adding them to a single relation
> (with the correct roles). For many people, creating a relation like
> this is seen as unnecessarily burdensome when for them, the
> addr:street method describes what they mean just fine.
>
> For data consumers (Nominatim, postcode finder, routers etc.) I would
> argue that it's conceptually simpler to use the relation method since
> it's explicit about that's in the street and what's not. With the
> addr:street method, the programmer has to, upon finding a node with an
> addr:street tag on it, search 'nearby' in the database. This means
> that they must have a georeferenced database set up and these querys
> are not as fast as simply node/way lookups. Also, to assemble a full
> street (from many ways and houses) would require an iterative building
> process. For me as a data consumer, the advantage of the
> associatedStreet relation comes from explicitness and speed.
>
> What I think we need (and what I've started to look at) is a plugin
> for JOSM (or built-in for Potlatch) which allows you to simply select
> the houses and the street, press a button and the relation is
> magically created. If some of the selected items are already in a
> relation then the rest of the items are added to it.
>
> Some of this is just my (biased) opinion so feel free to do whatever
> works best for you (of course, you could always do both). For
> information on exactly how I tag addresses with relations, see [2]
> (which I'd like to announce is up and running again).
>
> Cheers,
> Matt
>
> [1]
> https://wiki.openstreetmap.org/wiki/Proposed_features/House_numbers/Karlsruhe_Schema
> [2] http://milliams.dev.openstreetmap.org/postcodefinder/tagging/
>
> _______________________________________________
> Talk-GB mailing list
> Talk-GB at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-gb
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-gb/attachments/20121019/7cb010d4/attachment.html>