[OSM-talk] Should Bridges be independent of their ways?

[Anthony]

On Sun, Sep 20, 2009 at 12:49 AM, John Smith <deltafoxtrot256 at gmail.com>wrote:

> 2009/9/20 Anthony <osm at inbox.org>:
>
> > That's an editor issue.  If the editor wants to display lanes in a single
> > way as parallel ways, and let you edit them if need be, it can do that.
>
> It's also a DB/framework issue, I don't think relations should be
> abused for this purpose, instead the DB needs to be extended to cope
> with lanes being individually tagged.
>

In most cases I don't think a relation is the only solution either.  I don't
see it as an abuse, though.  It is clearly being used to show a relation
among multiple ways.  In some cases I can see a reason to tag subsections of
a way.  Different lane speeds in a way in which traffic is free to change
lanes would be one example (although I'd caution you to ensure that maxspeed
remains on the way as the maximum speed you can travel in *any* lane,
because that is how you calculate the cost that needs to be input into any
shortest path algorthim).  In many cases I see a reason to create multiple
ways, and in fact in many cases (including the example at the top of this
thread) that is what is currently being done.  If there is separate routing
information, you need separate ways.

As long as you only suggest tagging subsections of a way in those limited
cases where there is no routing information conveyed, I don't really care
how you implement it.

For example, you may have different maxpseed=* depending on the lane
> you are in, for example near here there is 2 lanes, one in each
> direction, each lane for a short section has a different maxspeed,
> I've taged it as maxspeed:forward=* but this won't scale to say 5
> lanes in each direction and 4 different maximum speeds depending on
> which lane you are in, and I know this situation exists as I've seen
> it. 130, 110, 100, 80, 80
>

If there are 2 lanes going in the same direction, with maxspeeds 130 and
110, the maxspeed in that direction is 130.  The fact that you have to be in
a particular lane to travel at that speed is irrelevant to routing software
so long as you can change lanes freely.

On the other hand, if you have to pick which lane to be in, and stay in that
lane for a certain significant period of time, then there should be a split
into two ways at the decision point, because routing software needs to know
about that - to say "get in the left lane if you want to go faster" or
whatever.  Do you see how if you are free to change lanes there's no need
for the routing software to tell you anything (and in fact, it *should not*
tell you anything)?  You just look up at the signs and make your own
decision, which you can change back at any time.

Per lane speeds, when you are free to change lanes, are fairly irrelevant.
I don't mind if you want to extend the system to handle them, but don't
suggest this extension has anything to do with other problems where we
already have multiple ways.

To wit, this thread started with someone talking about "a cycle lane, dual
carriage way (with central reservtion) & footpath".  Are you suggesting that
this should be represented as *a single way*?  If so, I have to vehemently
disagree.  If not, then tagging subsections of a single way has nothing to
do with the thread, and is off-topic.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20090920/8176c68f/attachment.html>