<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">On 2012-11-29 09:23, Jan-willem De
Bleser wrote :<br>
</div>
<blockquote
cite="mid:CAHgrPWLhgMCbHSrY2xJTGgHwy5hHdD81Da0AHODDPkTk1K45qg@mail.gmail.com"
type="cite">
<div class="gmail_extra">
<div class="gmail_quote">On Thu, Nov 29, 2012 at 4:51 AM,
A.Pirard.Papou <span dir="ltr"><<a moz-do-not-send="true"
href="mailto:A.Pirard.Papou@gmail.com" target="_blank">A.Pirard.Papou@gmail.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000">I have presented this
to tagging@osm, and I think I mentioned it on talk-be@osm:<br>
<br>
The municipality (L8=level 8) border segments (ways
between two municipalities) should be assembled with
multilinestring to form arrondissement L7 border segments.<br>
Then, the border of the arrondissement are now a much
smaller number of L7 segments.<br>
We may do the same at higher levels.<br>
The L8 borders are tagged admin_level=8,
name=municipalityA — municipalityB<br>
The L7 borders are tagged admin_level=7,
name=arrondissementA — arrondissementB<br>
The L6 borders are tagged admin_level=6, name=provinceA —
provinceB<br>
and so on for upper levels or lower levels if they exist.<br>
<br>
And then the meaningless saying "the highest admin_level
wins" goes away by itself, especially when applied to
names for which there is no reason to apply that rule.<br>
<br>
THAT is consistent, coherent, compatible, congruous,
harmonious, homogeneous, logical, solid, sound,
straightforward, uniform, you name it.<br>
<br>
But... no answer that proposition.<br>
</div>
</blockquote>
<div><br>
You're right, that does solve the "which layer?" problem. If
you mentioned that earlier in the thread, I'm sorry, I must
have missed it.<br>
<br>
The problem I have, however, is that by using name=A-B,
you're trying to give the boundaries a name when it really
is the municipalities that have a name.<br>
<br>
To use your example above, what if the L8 boundaries are all
members of multipolygon relations, each with the name of a
municipality, the L7 members of multipolygons named after
arrondissements, and so on. If you have the border, it is a
single api call to find which relations it is a member of,
and then you can easily extract the name. This is pretty
much what they suggest on the wiki (well, that or left: and
right: tags). I assume your program could do that extra
query without difficulty? Should be easy in Josm as it grabs
any relation in the bounding box, but I'll have to take a
look at Potlatch to see if it's possible there.<br>
<br>
Essentially, I don't want to have to "agree" on a name, I
want to use the one that's already there.<br>
</div>
</div>
</div>
</blockquote>
When I started to map Belgian boundaries, I looked for instructions
on the Belgian pages and I found none. So, I looked at what was
being done, I saw that names were used on boundaries and I did the
same. And now I am <i><b>accused of </b></i><i><b>insisting</b></i>
to put names on the borders when I found them that way.<br>
<br>
If I look at the result of the way it is done, I see nameA — nameB
name1 name2 name3 name4 ... everywhere, sometimes being a
municipality, sometimes being an arrondissement, sometimes being a
province etc... without any clue for the map reader to know which is
which. And Namur or Liège can be three different types.<br>
The result of my view of the Boundaries is that, instead of seeing
this on the border<br>
Liège — Namur Havelange Huy Namur Dinant Clavier Liège<br>
one would see<br>
Clavier — Havelange Huy — Dinant (arr.) Liège —Namur (prov.)
<br>
beside the unavoidable pile of shit.<br>
To this, I'm answered that the pile of shit is very well the way it
is. That the nec plus ultra is a renderer taking the name (but not
the admin_level!) off the municipality relation without the faintest
notion of what are the names to be used for distinguishing admin
levels. And one will certainly not miss the occasion to roll out
the refrain that I want to tag for the rendering.<br>
<br>
I, who would certainly be glad to map any community border like the
Quartier des Marolles, am accused of not considering them as
administrative borders because they can be used as postal addresses
and of not mapping level 10 names everywhere.<br>
<br>
I just read a question of someone wanting to navigate down the
boundary tree. I do it, but the answer he received is that it is
not possible. It goes on..<br>
<blockquote type="cite">
<table>
<tbody>
<tr>
<td class="answercell">
<div class="post-text">Basically everything is "free-form"
in OSM. There are conventions on tagging, but there is
no guarantee people will stick to them.</div>
</td>
</tr>
</tbody>
</table>
</blockquote>
My own opinion is indeed that it's difficult and unreliable to
obtain data from OSM.<br>
But, after reading for boundaries that one does it that way and the
other another way and even in Belgium that they are nor doing it the
way they say they must do it, I have serious doubts about existing
conventions too, conventions allowing to scan the tree here and not
there.<br>
<br>
The net result of this is that I'm losing my time writing messages
in hope of doing something right, that I hate doing things wrong and
that, in consequence I'm leaving the boundary business.<br>
<br>
I will finish as perfectly as possible, as I did before, what I have
promised to do.<br>
<br>
I started my boundary work by fixing borders that were 250 m away
from their position and putting Banneux that was in arr. Verviers in
arr. Liège.<br>
<br>
Don't forget to fix the other ugly, huge offsets in Belgian borders.<br>
<br>
Cheers, <br>
<br>
<table>
<tbody>
<tr>
<td valign="top">André.</td>
</tr>
</tbody>
</table>
<br>
<br>
<br>
</body>
</html>