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

[Anthony]

On Mon, Sep 21, 2009 at 6:20 AM, Richard Mann <
richard.mann.westoxford at googlemail.com> wrote:

> If we were just gathering data for routers, we would map every lane as a
> separate way, with relations for moving between each pair of adjacent lanes.
>

I disagree with that.  Dealing with relations when mapping a route is an
unnecessary burden.


> If we were just gathering data for rendering a single-scale street map,
> we'd add tags to a single way, and probably not bother with lane info.
>

Depends how micro you want to get.  With the detail people seem to want,
you'd probably be best off with a paint program :).


> I think routers ought to be able to cope with a single-way structure with
> extended tags. I think renderers would struggle (even more) to produce a
> good-looking multi-scale map if the data is held as multiple ways.
>

> I think the most generally-usable structure is a single way with lane info.
> This could be done with tags (and probably needs to be, to start with),
> though you might hope that editors would find a way of presenting it so that
> it looks like a set of parallel ways with tied nodes.
>

John seems to combine everything into a single way and treat the individual
"lanes" (some of the substructures aren't even really lanes) as
substructures.  Some people want to break every lane into a separate way,
and combine them into superstructures.  Frankly, these two plans are
essentially equivalent.

Personally, I don't want either of those.  I want the way to be whatever
logical unit is used for routing.  Then, you can take those ways and break
them into lanes (so long as the lanes do not have their own geometries
except maybe one node per lane at each endpoint).  *Or* you can combine
those ways into multiway structures like bridges.

As long as you keep all the detail provided by that, I don't really care how
you implement it, though.  If you implement it in a way which is burdensome
to routing software, the routing software can just reformat it before
processing, so long as you have all the data.  I think it's silly to combine
multiple ways into one and then create a structure to break them apart
again.  And I think it's silly to break ways apart and then create a
structure to put them back together again.  But if you want to go through
that process, so long as you keep the information that's currently there,
I'm not going to object.

I will object to any plan to combine multiple ways into one which *doesn't*
maintain the information provided by having multiple ways.  And the
pseudocode John last presented did that, plus my discussion with him
indicates that he doesn't understand the purpose of keeping ways separate in
the first place (the contradictory comments about a bridge being a single
way and a physically separated road being more than one way indicate this
clearly).



On Mon, Sep 21, 2009 at 8:21 AM, d f <fac63tempt at yahoo.com> wrote:

> Back to basics:
> Are we all agreed that, in principle, it would be better to be able to
> have a single bridge to carry multiple ways? <http://osm.org/go/eukOONRtk>
>

I'm not totally convinced that this isn't just a software issue.  If the
renderers drew the ways as the proper width, they could avoid drawing bridge
indicators in places where there isn't any space between the roadways.

It wouldn't hurt to add this information even if it's redundant.  But I'm
not convinced it isn't redundant.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20090921/5f42a018/attachment.html>