[Tagging] RFC on two proposals: Motorway indication; Expressway indication

[David ``Smith'']

On Fri, Jul 16, 2010 at 4:55 PM, Anthony <osm at inbox.org> wrote:
> On Fri, Jul 16, 2010 at 4:39 PM, David ``Smith'' <vidthekid at gmail.com>
> wrote:
>>
>> Add to that
>> grade_separated=* and you would indeed describe what is physically a
>> freeway/motorway.
>
> What's the point of grade_separated=*?  I thought that was covered by
> bridge/tunnel=yes and the lack of sharing of nodes at the point where two
> roads cross.

Technically you're right.  But practically, grade_separated=* offers a
/summary/ characteristic about the linear road.  A nice map will
display the difference between freeways (entirely grade-separated) and
expressways (partially or not at all grade-separated) by choosing
different line styles for each.  The line style is applied along a
linear path, whereas your suggested test (sharing of nodes or lack
thereof) only provides results at discreet points along the road.
That information can, theoretically, be then applied to the length of
the road by a kind of "nearest-determination" method, or by a
less-sophisticated per-way AND operation; but then on roads that have
partial grade separation — many expressways — you'd have a weird
dash-dot alternation between freeway and expressway classification.  A
useful map should only mark a road as a freeway where it consistently
has grade separation from every crossroad.  Not to mention, the test
you suggested would have to be performed by every application that
wants to know whether the road is (in summary) grade-separated or not;
in the case of renderers, if this analysis is implemented at all, the
mediocre result would almost certainly be generated as a kind of
pseudo-tag anyway.  We might as well just tag it from the start, using
human judgement.

-- 
David "Smith"
a.k.a. Vid the Kid
a.k.a. Bír'd'in

Does this font make me look fat?