User talk:RicoZ/Deprecate man made bridge
What is the recommended replacement?
Currently man_made=bridge is the approved and commonly used way to map the area of a bridge (in addition to adding bridge=yes tags to the ways that cross over the bridge). Are you proposing an alternative? --Jeisenbe (talk) 06:34, 15 September 2019 (UTC)
- At this point I am not proposing an alternative. I think man_made=bridge is a reasonable concept..note that this draft is in my "sandbox" for a reason. RicoZ (talk) 18:55, 17 September 2019 (UTC)
Why not covered=yes on ways under the man_made=bridge area?
If you concern is that it's difficult to render properly the ways that pass under a man_made=bridge, why not propose using covered=yes for this, as with ways that pass under buildings like roofs? This is already pretty commonly used. --Jeisenbe (talk) 06:33, 15 September 2019 (UTC)
- I think it is a nightmare to maintain such data in complex situations and mapping something that the renderer can easily determine on its own. RicoZ (talk) 18:59, 17 September 2019 (UTC)
Note
Basing data model of OSM on specific technical details of specific data consumer is absurd and unwelcome. Especially in case where what is blocked is a minor, dubious feature request of a minimal value with no one interested enough to even try working on it. Note also, that as far as I am concerned rendering of bridge outlines is not ideal, and can be improved but is OK (like 90% of the OSM Carto map style).
Note that objects with highway=path, location=underground, location=roof, surface=unpaved on roads and other have much more significant rendering issues in this specific map style. That does not mean that deprecating all this tags is a good idea Mateusz Konieczny (talk) 11:11, 15 September 2019 (UTC)
- I agree it is bad to model OSM features around bugs of the main renderer and I hope this proposal will never become reality. In fact I am strongly in favor of man_made=bridge except that I am realist. man_made=bridge is a very important and visible feature and if there is no prospect that the main renderer will render it correctly in foreseeable future I see it as considerable problem. In fact I am strongly in favor of man_made=bridge except that I am realist.
- At some point we have to decide ... do we map for the renderer, map to workaround renderer bugs or press something new?
- if the renderer is too stupid (or just can't keep up with development) it would be a possibility to rethink the old wisdom to avoid mapping for the renderer and instead explicitly foster renderer hints
- workaround renderer bugs, with tunnel or covered... may be the easiest solution in the short run but especially in complex urban environments I am afraid it will lead to data that becomes not maintainable very quickly
- something new (just a bunch of ideas at this point):
- cleanup OSM data so that layer=* will actually become more usable for data consumers (remove layer from objects where it does not belong, fix wrong layer/location/level on all kind of objects.. add checks to validators etc)
- more complex bridges are rarely really isolated from the roads "bellow", we might use concepts similar to or borrow something from indoor mapping