[Talk-us] Highway Shield Rendering

Phil! Gold phil_g at pobox.com
Tue Apr 3 17:06:30 BST 2012

* Chris Lawrence <lordsutch at gmail.com> [2012-04-03 10:21 -0400]:
> - Secondaries (network US:VA:secondary) don't seem to be rendering at
> all, and the fallback shields aren't showing up even where there are
> ref tags (just seems to be using Mapnik style).  Simple rule for VA:
> if the ref >= 600, or it has a letter in it, it's a secondary (except
> 785 and 895, which are signed primary).  1 <= 599 are primary.

When we looked at the database, we saw some secondary routes tagged as
US:VA and some as US:VA:Secondary.  Since there didn't seem to be any
overlap in the numbering, we chose to only look for the US:VA network and
render either a primary or secondary shield based on the number.  I assume
you live in Virginia.  What do you, as a resident, think of this rendering

Separately, Mapnik ought to be using the fallback shields when it doesn't
place one of our shields.  It might be getting confused by the presence of
the US:VA:secondary route even though there aren't any shields for it.
I'll look into it.

> - The US 460 business route doesn't seem to be getting shields.

We're looking for US Business routes under a network of US:US:Business.
It probably isn't tagged that way.  Once it is, it'll show up.

> Also, a more general comment - I think concurrencies might look better
> stacked vertically in some circumstances... you'd have to have some
> logic about the underlying direction of the way to make that happen,
> but vertical stacking would look nicer on N-S ways I think.

Someone else had a similar comment.  I'm pondering ways of matching the
major axis of the shield clustering to the general direction of the way.
I don't think I can get this perfect in all circumstances: without some
alteration to mapnik's code, I think the best I can do is to get the
overall orientation of a way in its entirety.  That will be good enough
for a lot of cases, but if a way has a north-south section, then a curve,
then an east-west section, it's probably going to come out with a diagonal
orientation.  I'm going to do some test renderings and see how good I can
make it look.

> I-26 in TN seems to be missing:
> http://elrond.aperiodic.net/shields/?zoom=11&lat=36.35713&lon=-82.42503&layers=B0

The route relation has "I 26" in the ref tag.  Once it's change to just
"26", it'll render properly (although it'll take until the next time we
run the cluster generating script after that change before it'll show up
in concurrencies).

> Similarly, while the 4-way US multiplex over the "old bridge" in
> Memphis is rendered fine, I-55 seems to be missing in both AR and TN
> (but is OK in MS):
> http://elrond.aperiodic.net/shields/?zoom=14&lat=35.1354&lon=-90.07954&layers=B0

It's the same thing.  The route relation for I-55 has "I 55" in the ref

> Looks great so far otherwise - keep up the good work!


...computer contrarian of the first order... / http://aperiodic.net/phil/
PGP: 026A27F2  print: D200 5BDB FC4B B24A 9248  9F7A 4322 2D22 026A 27F2
--- --
If you employ people as advisors, listen to their advice.
                       -- Evil Overlord's Handbook, entry 17
---- --- --

More information about the Talk-us mailing list