[Tagging] Non-orthogonal crossing=* tag proposals: crossing=marked/unmarked vs crossing:markings=yes/no

[Nick Bolten]

Hi everyone!

I have two proposals out regarding the crossing tag and how it is not
orthogonal, leading to all kinds of issues in mapping crossings and later
interpreting that data. As currently written, if both proposals were
accepted, crossing=traffic_signals/uncontrolled/unmarked would become two
tags: crossing=marked/unmarked and crossing:signals=yes/no.

Both Tobias and Masteuz have made an interesting suggestions about
crossing=marked/unmarked, which is that it still has the problem of
declaring that a crossing has a "type" (marked or unmarked) whereas it
could be considered another attribute, just like having traffic signals.

To give background, I initially chose crossing=marked/unmarked because (1)
both are in use in the wild, (2) the schema is equally non-ambiguous, and
(3) if I had to decide on the "type" of a crossing, I'd separate those with
no indication of their presence aside from regionally-varying conventions
(which is currently mapped as crossing=unmarked) from all the rest. But
point 3 isn't completely true: a crossing that has only signals but no
clear ground markings is less abstract/"fictitious" than a crossing
established solely by convention, with no infrastructure saying where to
cross.

In contrast, crossing:markings=yes/no would let us avoid making decisions
about the "type" of crossing entirely. If it were swapped out for the
crossing=marked/unmarked proposal, it would result in this schema for
crossings:

crossing=no (for crossings that should be specifically called out as not
doable/allowed)
crossing:markings=yes/no
crossing:signals=yes/no
crossing_ref=* (unchanged)

There has also been the suggestion that crossing=* could be left unchanged,
and these two new tags added as alternatives. I like that this potentially
avoids conflict and therefore makes it easier to start mapping this data
separately, but think it would result in competing schemas and redundant
data.

So, what are you thoughts? Is crossing:markings=yes/no better than
crossing=marked/unmarked? Are there any downsides/upsides I've missed? If
crossing:markings were preferable, what should happen to the crossing=* tag?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20190524/7ea5761b/attachment.html>