[Talk-GB] Schools Progress Tracker Update

Dan S danstowell+osm at gmail.com
Sun Jan 24 11:57:01 UTC 2016

2016-01-24 11:42 GMT+00:00 Ed Loach <edloach at gmail.com>:
> Stuart wrote:
>> 1 site, 2 schools:
>> • boundary has amenity=school
>> • buildings have school names & e.g. edubase tags. I used amenity=school for
>> the individual buildings though, as well as building=school. It should probably
>> only be building=school, really, as the site is the amenity. But this way it gets
>> picked up on the match tool.
>> • I would ideally like to have named the boundary e.g. “Hamstel Schools”
>> or “Chalkwell Schools” but haven’t as that will (for now) lead to a false “look at” flag.
> I think what you describe as what you’d ideally like to do is what I did in those examples I mentioned in my previous email (I can't remember though whether I used building=school or building=yes).
>> 1 school, 2 sites.
>> • I used the site relation, via JOSM. I believe that this is the correct way to do it. I
>> tagged the site relation with the edubase code and names, and the individual sites
>> with the names of e.g. “XX upper school” and “XX lower school”. However, these
>> didn’t get matched.
> The site relation page however
> http://wiki.openstreetmap.org/wiki/Relations/Proposed/Site#Proposal
> suggests it should be multipolygon and not site -
> "For example the tag amenity=school describes the perimeter of the school grounds, for schools with multiple sites the multipolygon relation can be used. Usage of a site relation is not appropriate here."

Hi - it's an interesting ambiguity between "multipolygon" and "site".
I actually think the thing you quote is a bit mis-worded, and what
they're trying to say (I'm inferring from the other sentences in the
wiki page...!) is that you should use "multipolygon" to aggregate
multiple buildings (for example) that sit within a single grounds,
whereas you should use "site" to aggregate multiple objects that are
more widely separated ("scattered throughout across the city" is the
wiki guidance).

This shows that OSM could perhaps live without the "site" relation if
people simply used multipolygons. However I think people tend to
assume multipolygons are quite localised, which probably makes a
difference to how they are rendered (e.g. one label for a whole
multipolygon, vs one label for each member of a site).

Anyone else got input on this? I might tweak the wiki, if it seems I'm
not in the wrong.


More information about the Talk-GB mailing list