I am having second thoughts on the colon separator for
role=north:unsigned. The colon separator seems to be more common in
keys, like lanes:forward=2, lanes:backward=2 etc. while the semicolon
or pipe seem to be more prevalent to separate values. The pipe
character seems to be more widely used when there is an ordered set of
elements, like lanes:maxspeed=40|60|60 to indicate speed limits for
lanes 1,2,3 respectively, whereas the semicolon seems to be used as a
more generic separator like destination=Salt Lake City;Reno. (Even
there you could argue that there is an ordering, the first element
would appear first on the sign, the second one below that.)
So I changed the wiki
http://wiki.openstreetmap.org/wiki/Highway_Directions_In_The_United_States
to reflect this and propose the semicolon approach:
role=north;unsigned.
OK?
On Sun, Dec 15, 2013 at 11:37 PM, Martijn van Exel m...@rtijn.org wrote:
Edited the page to clarify, but if you think it needs more discussion
I'm happy to do that as well!
Also I think the page could do with some clarifying diagrams at this
point...Anyone good at that?
On Sun, Dec 15, 2013 at 11:27 PM, Martijn van Exel m...@rtijn.org wrote:
James, all,
Work on JOSM is underway, and should be finished by the end of this week.
I don't think I fully understand what you're trying to convey about
the local/express lanes, but I think we should ensure that both JOSM
and iD support cardinal directions with any :extension.
I did make significant edits to the wiki page to capture the
discussion and move ambiguous parts out of the way, but the
north;south bit is not mine and I actually don't think it's a great
idea - can't we just have role=north being concurrent with the OSM way
direction? Or is that an oversimplification?
Martijn
On Sat, Dec 14, 2013 at 4:41 AM, James Mast rickmastfa...@hotmail.com
wrote:
Looks good to me Martin. I'm game with the role = north:unsigned tagging
for unsigned segments.
Now all we would need to do is get JOSM to show the cardinal directions the
same way in the relation editor like forward/backward so that you can
verify a route is all there and there are no gaps (unless there is one for
real like I-49 currently has in LA since they are extending it). And on
this subject it brings up an interesting problem. What to do when a highway
has C/D lanes that are part of the main highway (like the 401 in Toronto,
Ontario, Canada). I know a few Interstates have these, like I-80 I-95 in
NJ. There should be a way to have something like role = east:express
role = east:local in a directional relation (I fully support Interstates
to have separate relations for each direction on 2di's; but on 3di's they
should stay one relation unless it's like a 30+ mile route like I-476/I-376
here in PA) and have JOSM's relation editor show a split in the highway so
you can verify there are no gaps in those areas for the relation.
Also, I have noticed you've been doing some editing for the Highway
Directions In The United States wiki page [1] and mention the role =
north;south idea for single carriageways so that the routes could tell
people which direction the way goes. I think that might still need a little
more discussion here on [talk-us] before we attempt to implement it and
mention it on that page (maybe have a vote for that part on the talk
page??). I personally think it could work, but we would need all of the
editors (JOSM, iD, Potlatch2) to have support to be able to reverse those
roles correctly if somebody reverses the way. Can't allow those to get
messed up once added. (On a side note, iD doesn't alert you if you delete a
way that's part of a relation yet, which isn't good at all.)
-James
From: m...@rtijn.org
Date: Tue, 10 Dec 2013 18:16:54 -0800
To: rickmastfa...@hotmail.com
CC: talk-us@openstreetmap.org
Subject: Re: [Talk-us] Separate relations for each direction of US State
highways.
Hmm yes, on second thought, a second key on role members may not be so
straightforward ;) How silly of me to suggest such a thing.
Let's keep things pragmatic then and let me suggest we go with
role=north:unsigned for unsigned sections. I don't particularly like
the ; because it suggests a list of things that are of similar nature
(like apple;pear;mango) whereas a colon to me suggests a further
scoping which is what this is.
So
role=north / role=west / role=south / role=east
for relation members to indicate cardinal directions, and
role=north:unsigned / role=west:unsigned / role=south:unsigned /
role=east:unsigned
for unsigned segments, unless the entire numbered route is unsigned,
in which case unsigned_ref would do the job.
Any more insights and comments?
Thanks
Martijn
On Fri, Dec 6, 2013 at 5:31 PM, James Mast rickmastfa...@hotmail.com
wrote:
Well, to add a second role to an item in a relation would require an
entire
overhaul of relations, the editors, and even the OSM database I would
