[Talk-us] Proposal: Sunset ref=* on ways in, favor of relations

stevea steveaOSM at softworkers.com
Wed Nov 11 19:38:33 UTC 2015

>On Wed, Nov 11, 2015 at 7:55 AM, Richard Welty 
><<mailto:rwelty at averillpark.net>rwelty at averillpark.net> wrote:
>and for unsigned routes, either don't create the relation, or create a
>relation and use unsigned_ref instead of ref

And Paul Johnson <baloo at ursamundi.org> replied

>I'm a proponent of preserving this data in unsigned_ref=* myself...

+1.  Please promulgate data so OSM does not LOSE or MISS route 
relation data if there is no sign, let's CAPTURE them with 
unsigned_ref=* if/as/when we can.  People might throw rocks at me for 
saying this, but when/as we get to decent shield rendering, my 
personal belief (contrary to Richard) is that unsigned_ref should 
render with admin-level appropriate shields, rather than NOT render 
because they are unsigned.  (We are talking about US-centric highway 
shield rendering renderers).  I don't think I'm alone here.  While 
"OSM purity" (via our "on the ground observable" tenet) might lean in 
Richard's direction, a data consumer looking for the right highway to 
travel on very likely wants to see a shield on a road map, even if it 
doesn't show up on a signpost on the road.  In the interests of 
publishing a more useful product better tailored to this specific 
data consumer, we can cut ourselves a little slack here by rendering 
an unsigned_ref=* as a shield.

I very much like these encouraging indications on this thread:

1)  Kevin Kenny <kennydb at acm.org> shared his "catskills" 
hiking-appropriate renderer,

2)  Paul points out that cycle and bus route relations already render 
with "OSM trade dress" renderers (those available from the Main Page),

3)  Paul Norman <<mailto:penorman at mac.com>penorman at mac.com> wrote 
Phil! Gold's demo is "an excellent proof of concept" (I agree) but it 
needs patching (or must use a modified Mapnik) to modify one or more 
read-write database calls to become read-only, AND it appears to 
remain needing to be ported to the new Carto stylesheets.  Much has 
been written in github's openstreetmap-carto/issues/508 and 
openstreetmap-carto/issues/596 about this, though with much being 
left unresolved or difficult to currently resolve.  These are 
promising developments, but there are serious technical hurdles.  I 
offer a genuine tip of my hat to excellent discussion on these github 
discussion threads.

4)  Richard writes (and others seem to concur) that Phil!'s demo does 
federal and state level shields well enough, but leaves county and 
city routes incomplete.  I find this a satisfactory place to accept a 
preliminary implementation, with lower admin-level (county and city) 
shields being completed over a longer time frame.  (Tough nuts take 
longer to crack).

Especially to the dozen-or-more core people who really know what you 
are talking about (you know who you are, alas, I am not one of them), 
please let the good discussion momentum about rendering shields (at 
least in the US) continue.  OSM as a project seems to have a fierce 
determination do this (and do it right and well), but only as much 
hard work and toolchain development continue.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20151111/364e3f81/attachment.html>

More information about the Talk-us mailing list