On 2018-09-03 21:59, Nathan Monfils wrote:
Op ma 3 sep. 2018 om 20:03 schreef André Pirard <a.pirard.pa...@gmail.com>:
Op zo 2 sep. 2018 om 22:54 schreef Nathan Monfils
<nathanmonf...@gmail.com
Hello!
I was doing some editing on the N30 near Liège, when I noticed that its
relation was completely unordered, with ways near the city center being
next
to ways much further south (see relation `124374`).
The wiki defines a relation as an *ordered list*, yet only mentions order
for
bus lines. Is there a need for the road relations to be ordered
correctly?
Regards,
Nathan Monfils
On 2018-09-03 00:24, Jo wrote:
for route=road relations order doesn't matter much. It's impossible to
sort them according to any meaningful criterion.
Op ma 3 sep. 2018 om 20:03 schreef André Pirard <a.pirard.pa...@gmail.com>:
A meaningful criterion to sort a route is so that two adjacent ways of it
share one same end node.
In other words, that the route is ordered the way you travel it without
interruptions.
That's what the "sort" buttons of JOSM do.
And, beside finding a segment more easily, the schema of the route made by
JOSM shows the gaps (missing pieces).
And in particular it will show if the road is circular.
For example, if you sort the N30, you will see a gap between Boulevard de
Fraipont and Avenue de la Libération.
And 16 other gaps in total.
Last time I spoke of routes with a Potlatch user, he told me he couldn't
sort routes.
I don't know about ID and that's quite a time ago.
The N30 should be sorted and corrected.
I will let Nathan do it if he's busy with it.
Amitiés,
André.
Polyglot
On Monday, September 3, 2018 8:09:28 PM CEST Jo wrote:
The "problem" with N roads is that they are not linear features, they
split, recombine, have dangling dead ends, roundabouts and so on.
Yes, you can group some of the elements, but the next time you sort, other
groups may be formed, so it's arbitrary.
Jo
I fixed the N30 relation the way I described.
I chose the order from Liège to Ardennes (holiday departures are happier
than comeback ;-)).
That's the element order, down the list in JOSM relation editor.
Roundabouts are indicated by role "forward".
First way set to follow when traveling in route's direction.
Second way set to follow in the other direction.
JOSM shows them nicely with sidings like this
splits/recombines like Boulevard de Beaufraipont are normally like
roundabout.
Basically, global sort doesn't change my roundabouts and splits.
But sort only sorts what is sortable and is only guaranteed as a local tool.
Hence, a global sort doesn't find a common northern node for the huge split
starting at Boulevard Raymond Pointcarré and ending in prongs.
And so it includes only one branch of it and it throws the other one to
the end.
All the best,
André.
On 2018-09-03 21:59, Nathan Monfils wrote:
After looking into it further, I can only agree. I added the missing members
back (one way and a few junctions that were probably overlooked by previous
mappers), but every roundabout and double road "cuts off" the automatic
ordering by JOSM.
It seems to consider that a loop breaks the relation order entirely and
doesn't attempt to follow it any further.
I still pushed the changes since they contain a few added members, but I think
the N30 (and other N roads) will have to stay unsorted, unless JOSM's sorting
mechanism starts taking geographical proximity into account instead of
strictly checking the end nodes.
Regards,
Nathan Monfils
_______________________________________________
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be
