[OSM-talk] Maximum recommended length of ways tagged with layer

Richard Z. ricoz.osm at gmail.com
Sun Mar 16 10:44:55 UTC 2014


On Sat, Mar 15, 2014 at 06:12:10PM +0000, moltonel 3x Combo wrote:
> On 15/03/2014, Fernando Trebien <fernando.trebien at gmail.com> wrote:
> > I now agree that the "layer" tag should be used as locally as
> > possible, so I think Richard had good intentions when proposing this.
> > At the same time, I think you, Frederik, has a good point that
> > arriving at a threshold for that number is quite hard. What exactly do
> > we want to avoid? Really, really long ways with a layer tag. So why
> > not set this threshold higher? Say 10 km?
> 
> Validator rules are a good thing, but I think that "length of a way
> that has layer=*" to detect misuse of the layer tag is beside the
> point. Whatever threshold you use, there'll false-positives and
> false-negatives. How about something along the lines of "negative
> layer but no tunnel tag (or positive/bridge) and no/too many crossing
> ways" ?

I am now thinking about

"unless absolutely necessary the size of objects tagged with a layer tag
 should  not exceed a size which would be typically downloaded for editing
 in this area."

but the wiki page already says 

<<
* Tag shortest possible/practical sections of ways. Long viaducts and tunnels 
  can be tagged with a suitable single value for their entire length for simplicity 
  although it may sometimes be better to adjust the layer along its length to accommodate
   more complicated crossings.
* Use the smallest suitable layer value. Only use layer=2 for a bridge that passes 
  over a feature that is already at level 1; similarly only use layer=-2 for 
  a tunnel that passes below another tunnel. For convenience some higher values 
  are often locally used/reserved for very long bridges or underground networks 
  where it is assumed that they are above/bellow most other crossings/objects in the area. 
>>

- which should be good enough if people don't interpret the text in some 
unforseen way.

Richard



More information about the talk mailing list