Proposal talk:Default layer for bridge and tunnel

We should not substitute explicit data with assumptions

This change serves no purpose (laziness was given as a reason on russian forum, but that's out of OSM scope and should not be fixed by cripping data scheme), while bringing many problems:

• First of all, your understanding of layer tag in incorrect. The tag, as stated in wiki, is used to describe relative position of objects - e.g. layer only has meaning in presence of other objects at different levels. Given that, you cannot globally assume layer for any object, not knowing layers of objects around it. Also, we should not have tags that change values of other tags or tags that change their default value depending on whatever.
  • But ground level is tied to 0 layer. - Vort 11:47, 15 December 2010 (UTC)
  • NACK (and agree with Vort). If one high/waterway crosses another high/waterway, then by default the one with the brige-tag should be above (=+1) the one without (=0). If one high/waterway crosses another high/waterway, then the one with the tunnel-tag should be below (=-1) the one without (=0). There is no need for an additional layer tag in /those/ cases. In any other case (eg. two bridges or two tunnels crossing) needs layer-tags. /al 10:54, 17 January 2011 (UTC)
• There are cases where 1/-1 assumptions are just incorrect, e.g. flat bridge over river should more likely be layer=0 with layer=-1 on river to show that river is lower than the "ground level" (http://upload.wikimedia.org/wikipedia/commons/4/40/Bad_III.JPG).
  • Quote: "If the default value of layer=-1 is not suitable (there are already layers), add layer=0/-2...-4 or even 1...4." There is also in view of the surrounding area. --Canabis 11:08, 15 December 2010 (UTC)
    • Adding more conditions to this won't make this more useful. We should just map level unconditionally to avoid any errors and ambiguities.
  • layer 0 means that object lies on the ground. In this case objects are layered as follows: ground, water, air, bridge. Water is on the ground, not underground, so tagging river with layer = -1 is incorrect. - Vort 11:45, 15 December 2010 (UTC)
    • layer=-1 is more correct (because it is canal) than layer=1 for environment. --Canabis 12:11, 15 December 2010 (UTC)
    • It is still below ground level and all objects "on the ground" --AMDmi3 12:27, 15 December 2010 (UTC)
      • If you dig a hole in the ground, it's surface will not become a layer = -1. There is no difference between natural and human-made holes. - Vort 13:10, 15 December 2010 (UTC)
      • I made a mistake. Corrected "then" -> "than". --Canabis 13:25, 16 December 2010 (UTC)
• It automatically adds errors on the map, e.g. if there are two crossing bridges one with layer=1 and another without (correct case at present), now they're crossing in midair. Same for tunnels.
  • Show examples please. - Vort 11:45, 15 December 2010 (UTC)
• It guarantees that more errors will be added in the future, e.g. when newbies who look at the map to see how stuff is tagged will get the idea that bridge/tunnel does not need layer at all, and will not use the tag when it's needed
• This is extra complication in documentation (layer tag description), code (validators and renderers code which now need to check for extra assumptions) and minds (people will have to rememberer all the conditions)
• How to go from scheme used now to new scheme? How to understand if now no layer - is author means that layer=0 or miss them? Dkiselev 12:38, 15 December 2010 (UTC)
  • We can set layer=0 to all nowaday objects without layer tag. --Canabis 13:25, 16 December 2010 (UTC)
• I think, it's better to warn mapper if no layer tag on bridge/tunel, in validators Dkiselev 12:45, 15 December 2010 (UTC)
  • In JOSM since 2010-11-25 Validator integrated in kernel, but mistakes continues. About other editors, I don't know. Examples for December: http://osm.org/go/ZGU6eNBpU-, http://osm.org/go/ZGeC5jPtY--, http://osm.org/go/ZSCPWJPxd-- and others. --Canabis 13:27, 16 December 2010 (UTC)

Happy mapping. http://upload.wikimedia.org/wikipedia/commons/2/27/Passerelle-tolbiac-ouverture-pano.jpg

--AMDmi3 12:27, 15 December 2010 (UTC)

All the given examples above can be covered by explicitly defining level=*. Around 95% of the tagged bridges/tunnels are level=1/-1. Untagged is level=0, as far as I know, which probably is in 95% of the cases wrong. Why not correcting the majority of these errors by setting a default level for bridges and tunnels? Of course this may introduce some other errors, but much less then it solves. --Teddych 20:30, 15 December 2010 (UTC)

• Why not correct these errors by correcting these errors (by adding level where it's needed) instead of adding complication that adds more errors on its own? --AMDmi3 09:21, 16 December 2010 (UTC)
  • To avoid redundancy. - Vort 09:38, 16 December 2010 (UTC)
    • There is no redundancy to avoid --AMDmi3 10:29, 16 December 2010 (UTC)
      • At the moment nearly not, you are right. With level=1/-1 as default we would have 95% of the bridges/tunnels mapped correct without a redundant level=1/-1 tag. A lot of correction work could be avoided with that. --Teddych 14:52, 16 December 2010 (UTC)
        • Not avoided but hidden. Errors (no layer for overlapping objects, which is required data in the database, not closely a redundancy) will remain and there will be even more of it. --AMDmi3 16:38, 16 December 2010 (UTC)
          • Defining a default layer is implicit data in the database. The discussion is not about removing data (removing layer), it is about redefining the default that would reduce incorrect data in the database. --Teddych 07:04, 17 December 2010 (UTC)
            • It is all linked. "No need to" add data -> it won't be added, all the errors remain. Also, there were already suggestions to remove level=1 from bridges "in the future". --AMDmi3 09:55, 17 December 2010 (UTC)
  • Because mistakes continues. Examples for December: http://osm.org/go/ZGU6eNBpU-, http://osm.org/go/ZGeC5jPtY--, http://osm.org/go/ZSCPWJPxd-- and others. --Canabis 13:27, 16 December 2010 (UTC)

Looking at the most simple case, two highways crossing with one tagged bridge=yes, there's absolutely no reason I can think of that the bridge would not be on top. Therefore explicitly tagging the bridge as layer=1 serves no purpose other than making up for flaws in the renderer. --NE2 23:31, 16 December 2010 (UTC)

What about...

How many of those bridges without a layer tag (so currently default to layer=0) are crossing ways at layer=-1? I've seen many places where bridges cross rivers or rails which have been at layer=-1 (though to my way of looking at it, they should have been zero and the bridge at 1).

By adding a default layer=1 to a bridge, this would not affect the concrete example. layer=1 of bridge is higher the rail/river and this is important. --Teddych 06:58, 17 December 2010 (UTC)

So in my opinion it is best to leave the default as that which the majority of active mappers know and correct the bridges that need correcting rather than try changing the default.

Although feel free to make any assumptions you like in your own consumption of the data (rendering or whatever).

-- EdLoach 15:24, 16 December 2010 (UTC)

• It's correctly when bridge with layer=1 as default will be crossing ways at layer=-1. At least not an error or ambiguity. --Canabis 11:06, 17 December 2010 (UTC)

Meaningless

/discuss. --Hind 11:54, 11 January 2011 (UTC)

So true. And where are the RFC and vote start messages sent to the appropriate tagging mailing list? Alv 15:20, 14 January 2011 (UTC)

This one is just plain going to fail. It's 5 for, 15 against, and 1 neutral. -- rickmastfan67 12:34, 15 January 2011 (UTC)

I was sent messages to tagging(AT)openstreetmap.org 15.12.2010 about start RFC and 11.01.2011 about start voting. Messages had to go through after being moderated because I am not a member of the mailing list. I didn't recieved notification of block of messages. --Canabis 12:54, 15 January 2011 (UTC)

They haven't been let through, there's probably too much spam that anybody with rights would even look at the messages that need moderation: December, January. It's even more unlikely that rejected authors would get a notification, as they are mostly automatic spammers. Alv 13:24, 15 January 2011 (UTC)

I relied on the phrase in the auto answer: "Either the message will get posted to the list, or you will receive notification of the moderator's decision." I would get registered in the mailing list and tell there about voting. Besides the reference to the proposal was in the category "See also" of many related wiki-pages.