[Tagging] Deprecation - waterway=riverbank vs water=river

[Kevin Kenny]

On Wed, Feb 10, 2021 at 10:05 AM Tomas Straupis <tomasstraupis at gmail.com>
wrote:

>   Basically what I've already mentioned:
>   1. QA tools: there are numerous QA tools (not only keepright,
> osmose, geofabrik) which are doing periodic data verification,
> riverbank is a very important feature there it is included in many
> rules. Those would be topology rules (riverbank must not overlap other
> first grade features, waterway must have waterway=river|stream
> crossing). QA tools help detect accidental damage of data by newbies
> or people misusing tools, it also is essential for cartography: some
> calculations require some specific rules to be true.
>   2. Cartography - you display rivers as ways at some scales, but
> polygons (riverbanks) on larger scales.
>   3. Videos/blogposts/training material/books. These are especially
> hard because it could be minimal mentioning of the tag in some
> example. It would be very hard (say videos) or impossible (printed
> books) to change to the new version.
>

Putting on my data consumer's hat: Anything I do involving rivers has to
consume both waterway=riverbank and natural=water water=river, and to treat
them as synonymous. I grumble a bit at waterway=riverbank because its
topology is different from natural=water, largely because it appears to
have been invented before multipolygon relations, but as a practical
matter, I know that I'll have to recognize both forever. I agree with you
that waterway=river is important (should follow approximately the Thalweg,
and should connect downstream to something). Since connectivity requires
that the Thalweg continue through lakes (which are already natural=water),
once again I don't really care what sort of water surrounds the
waterway=river.  Similarly, with cartography, you display the Thalweg as a
linear feature at some scales, and expand at larger scales to drawing its
banks, whether they be 'natural=water' or 'waterway=riverbank'. It's really
a non-issue.

Putting on my programmer's/data modeler's hat: I have a weak preference for
the 'natural=water' scheme because the topology is the same as any other
multipolygon. But if the other way is what''s specified for good reason, I
can implement it. In this case the good reason is that data in both
representations are already in OSM, and I have to deal with them.

Putting on my mapper's hat: Often, when I map either, someone comes along,
changes it to the other, and scolds me in a changeset comment. I act
contrite and leave it the way that the other person put it, as long as the
topology remains correct for the method chosen.

Putting on my prognosticator's hat: I foresee that OSM or a successor
project that builds on it will come to evolve into a more formal schema
with less of a 'folksonomy' model. 'Any tag you like' works when you're
doing something for the first time, but works a lot less well when you're
mapping a common type of feature and want what you map to be useful to
others. But I don't expect to see that process converge - particularly for
recurring contentious areas such as forests and riverbanks -  in my
lifetime. I'm an old man. In the meantime, I try not to spill the tag soup,
and to mop it up when I do.

-- 
73 de ke9tv/2, Kevin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20210210/03dacdd0/attachment.htm>