[OSM-talk-be] Some information about Nominatim for addresses in Brussels

[Marc Gemis]

The order on name can be random. But once you chose e.g. FR - NL, all
the addr:street tags for the addresses in that street have to follow
FR - NL. For another street you can take another order.

This weird rule of FR - NL or NL - FR as name is something from the
Belgium community. You cannot expect that all tools honour this. How
can a tool know that in this small place on earth the name field is a
concatenation of two names? By using FR - NL as name, you tell all the
tools that the name of the street is "FR - NL". So you have to use
that name everywhere you refer to the name of the street, which means
the field addr:street has to be "FR - NL" as well.
You do not use different names for addr:street and name for roads in
Flanders, do you ?

IMHO, this is not mapping for the tool, but is mapping in a consistent
way. Once the order is determined for a street, we also agreed not to
change it, so once cleaned up, there is no longer an issue.

But if people don't care about the first impression people get when
looking up an address in Brussels, fine for me.
But it is already bad enough that when you look for Zutphen in a Dutch
browser, you find "Zütphen".
All those little things might turn people away from OSM... :-(

m.

p.s. alternatively, you could use an associatedStreet-relation to
solve this problem. But this is not one of lonvia's preferred
solutions.






On Wed, Sep 28, 2016 at 10:18 AM, joost schouppe
<joost.schouppe at gmail.com> wrote:
> I'm not sure I understand correctly, but isn't the order of languages in the
> general name tags supposed to be random (for political reasons)? If so,
> shouldn't Nominatim be able to deal with this randomness?
> It sounds a bit strange to expect the data to follow the same randomness in
> the name and addr:street for different objects. Then upon every edit of a
> new object, you would have to check which randomly selected order of names
> is present on the street itself.
>
> To me, it looks a bit like "mapping for the tool" if we do clean this up.
> But I might be missing the point.
>
>
>
> _______________________________________________
> Talk-be mailing list
> Talk-be at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
>