[Tagging] Bridges and layers

[Richard Mann]

Dave F (et al),

Renderers draw roads (typically) by drawing a wide grey line on each
segment, a grey circle at each node, then a narrower (say) white line
on each segment, and a white circle at each node. All you see of the
grey is a thin line on each side of the white line: this is the
casing. The circles on each node are called caps (end-caps if they are
at the end of a way, join-caps if they are at an intermediate node).

If you don't draw the circles on the nodes then you get gaps at
corners. If you draw all the grey first, regardless of layer, you get
gaps in the casings when one road goes over another, so it looks like
they join when they don't (which happens on cyclemap). If you draw the
grey in the correct layer, then you get little semi-circular arcs of
grey at the end of bridges (if they are layer=1).

So renderers have to do something. Different renderers have come up
with different solutions, but all produce artefacts because there's a
piece of data missing (which they could pre-process, sure, but there
are better uses of time). It would be more effective to give them the
data, and have renderers do it reasonably well consistently.

What I've suggested isn't the only solution, but it's the most
economical, I think.

Richard