<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">On 2013-03-24 18:32, Frédéric Rodrigo
wrote :<br>
</div>
<blockquote cite="mid:514F38C2.2050303@gmail.com" type="cite">>
It makes no sense.
<br>
> route=bicycle is a tag for a relation, not for a way.
<br>
> It's very surprising that Osmose commands to repeat such a
tag on
<br>
> ever cycling way !!!
<br>
> How could we identify a cycling route if all the cycling ways
<br>
> contained route=bicycle ????????
<br>
<br>
> Osmose seems to be "rather stupid", and decides that if x
number of
<br>
> objects has certain tag combinations, all objects that have
one of
<br>
> the tags, should also have the others. It does not verify if
the tag
<br>
> is appropriate for the object (node, way, relation). But it
is still
<br>
> a valuable tool.
<br>
<br>
It's an automatic analysis based on statistical and an arbitrary
trigger at 50 count. I can change the tigger, eg with relative
value.
<br>
<br>
Error was triggered with a "parent" tag like in
highway=construction and construction=primary.
<br>
<br>
The analysis have already a tag blacklist. I can add
route=bicycle.
<br>
<br>
> These tags were used on some ways for some routes: if a route
goes
<br>
> over a big square, an additional way is created going across
the
<br>
> square.
<br>
<br>
Really need tag on this ways ? Or add the area to route relation,
or just no way at all.
<br>
<br>
> and I guess osmose wouldn't like untagged ways either
<br>
<br>
Osmose report error on untagged way not part of relation (with
role).
<br>
<br>
> But I haven't found any ways with both route=bicycle and
bicycle=yes
<br>
> though.
<br>
<br>
I found lot of them here :
<br>
<a class="moz-txt-link-freetext" href="http://overpass-turbo.eu/?Q=way(%7B%7Bbbox%7D%7D)%5B%22route%22%3D%22bicycle%22%5D%5B%22bicycle%22%3D%22yes%22%5D%3B(._%3B%3E%3B)%3Bout%20meta%3B%0A&C=50.67634;4.60636;12&R">http://overpass-turbo.eu/?Q=way(%7B%7Bbbox%7D%7D)%5B%22route%22%3D%22bicycle%22%5D%5B%22bicycle%22%3D%22yes%22%5D%3B(._%3B%3E%3B)%3Bout%20meta%3B%0A&C=50.67634;4.60636;12&R</a>
<br>
</blockquote>
<br>
<br>
Frédéric, my text you quote above has been replaced by this:<br>
<blockquote type="cite">
<blockquote type="cite">I completely change my mind.<br>
route= on a way can be ambiguous (in case of multiple routes
knowing and where they go at junctions) and the better place for
it is in a relation, in which case it's optional and not very
useful on the way.<br>
<b>But it is not invalid and there's no compelling reason to
remove them.</b><b><br>
</b> <b><br>
</b><b> What is incredibly surprising is that Osmose commands to
add route= on ways where there is no route.</b><b><br>
</b><b> Osmose should remove those warning urgently.</b><br>
</blockquote>
</blockquote>
<br>
<blockquote cite="mid:514F38C2.2050303@gmail.com" type="cite">There
is some choses:
<br>
- change the data
<br>
- black list route=bicycle
<br>
- use relative value on analyser to trigger at higher count.
<br>
</blockquote>
<br>
You should obviously blacklist asap.<br>
But you should also refrain from using statistical methods to invent
non existing OSM rules.<br>
<br>
Cheers, <br>
<br>
<table>
<tbody>
<tr>
<td valign="top">André.</td>
</tr>
</tbody>
</table>
<br>
<br>
<br>
<br>
<br>
</body>
</html>