[Tagging] Naming boundary ways - the — separation character ... or without any

A.Pirard A.Pirard.Papou at gmail.com
Wed Oct 10 23:07:53 BST 2012


PS:  I notice before sending that replying in 2 lines is faster ;-)

On 2012-10-10 21:26,  Martin Koppenhoefer wrote :
> 2012/10/10 sly (sylvain letuffe) <liste at letuffe.org>:
>> words) on which he replied that I was the only one arguing against 
>> and that
>> every one was happy with a — instead of more common characters like / 
>> - or
>> whatever.
>
> I think that "-" is not a good choice because there are some places
> that already have the hyphen in their official name, like
> Castrop-Rauxel or Dessau-Roßlau for instance.
> http://en.wikipedia.org/wiki/Castrop-Rauxel
Albeit A - B is not the same as A-B, I agree and that's most certainly 
the reason behind —.
Agreeing on a distinctive character is giving parsing by software a chance.
But — on a map needs very good resolution and font to avoid human 
confusion with -.
> There is also a common use of "/" to distinguish places (e.g.
> Frankfurt/Main, Schwedt/Oder, where Main and Oder are rivers used to
> distinguish the cities/towns from others, but maybe we could use the
> also common Frankfurt(Main) and Schwedt(Oder) instead, not sure if we
> need this at all, I think the official name would be "Frankfurt am
> Main" or "Schwedt an der Oder" and the above citations are short forms
> of it).

Invention has no limits:  Saint-Georges-s/Meuse = 
Saint-Georges-sur-Meuse ("sur" = your "am").

And "/" has just been used in   Deutschland — Belgique / België / Belgien
Fortunately, Belgium has no border with Switzerland ;-)  (1)

Well, my personal conclusion is that we should use NO separation character.
We should simply write the two parts facing each other on each side of 
the boundary (1).

1) For type=boundary, the name in the relation is single and the 
boundary is a loop.
So, it's nothing else that our request to renderers to determine the 
inside and write the name there (1).
Having it face the opposite name(s) would be a bonus raising our kudos.

2) For boundary=* & type!=boundary, I think I've read that the 
right:*=*/left:*=* have been deprecated because the *= were country 
specific (and admin_level is not, well, less).
But why not a rendered, plain left:name=*, right:name=*?
Up to the "level" stuff to identify what "name" is.

However, I (already wrote I) don't like those left: and right:.
It's just too easy for someone with just his landuse or hiking in mind 
to revert the direction of a border and cause thousands and thousands of 
people to have to move to where their country has gone ;-)
I had suggested N/E/S/W but I admit, it's better but not fail-proof.
A better idea would be to define the direction itself in an arrow-less, 
irreversible manner.
Something like N->S but E->W when the way slopes < 45° end to end, and 
clockwise when it loops.
Not bad, but 1) how would we declare the new scheme?  2) ouch: splits 
(again).

x) I recommend writing the level on manifold borders :
Liège - Verviers (arr.)    Liège - Luxembourg (prov.)  (1)
Liège is 1) a town,  2) a commune,  3) an arrondissement,  4) a province.

Cordialement,

André.


(1) It sometimes helps helping the renderer.
This is what Mapnik presently writes right on the ridge of northern 
France-Wallonia border:

France métropolitaine   Bachy   Tournai   France — Belgique / België / 
Belgien   Lille   Nord   Tournai-Hainaut   Nord Pas-de-Calais   Rumes   
Communauté française   Mouchin   Hainaut   Rumegies  Wallonie   
France-Belgique   Parc naturel régional Scarpe Escaut

Convincingly not in Russia.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20121011/1d51f718/attachment.html>


More information about the Tagging mailing list