[Tagging] Stop the large feature madness (was: Tag for a plateau or tableland?)

Kevin Kenny kevin.b.kenny at gmail.com
Thu Apr 18 02:47:18 UTC 2019


On Wed, Apr 17, 2019 at 7:55 AM Christoph Hormann <osm at imagico.de> wrote:
> As a rule of thumb i'd say something that can at least coarsely be
> surveyed on the ground by a single mapper during a single day is
> usually suitable to be mapped as a distinct named feature, provided it
> is otherwise verifiable of course.  For larger things mapping should
> focus on locally mapping locally surveyable constituent parts or
> aspects of the feature but i would be very careful with creating
> features for them as a whole because this very often drifts from the
> OSM idea of mapping local knowledge to a Wikipedia-like recording of
> social conventions.

I doubt very much that you're saying what you intended here.

It comes across as saying, for instance, that lakes too big to map on
the ground in a single day should not be mapped, or should not be
named. I think that making large waterbodies disappear would be
ridiculous.

Moreover, if you've mapped something on the ground, what difference
does it make how long it took?  It took me a number of trips over many
days to gather the GPS tracks that were consolidated into
https://www.openstreetmap.org/relation/4286650. (It was initially
planned as a single trip with two stops for resupply, but the best
laid plans gang aft agley.) On one of those trips I was in the field
for six days straight, and was at times thirty km from the nearest
drivable road. Of course the relation has many constituent ways,
because of tagging for things like bridge=yes or ford=yes, changes of
surface=*, brief stretches where the trail follows a road, and similar
changes. Moreover, I intentionally broke the ways up so as not to have
thousands of nodes on any signle way. But on the ground, the relation
represents a single trail. It has the same name for its entire length
(and is signed where it shares the way with a highway). Is it less
worthy of mapping because in order to order to map one section, I had
to lug enough batteries to keep my GPS going (and enough food to keep
Kevin going) for six days?

Surely you're not arguing that I can't have 'local knowledge' of it
when I've personally had my literal boots on the literal ground for
every step of the way?

I understand that there are fairly severe technological issues at
present, where a plethora of enormous multipolygons breaks some of the
software tools. For now, therefore, I refrain from mapping anything
like the Long Island Sound or the Red Sea as areas, even though I
believe that competent label placement in some renderings will require
that eventually. Similarly, I'm not about to go mapping enormous
linear features or area features for the Mogollon Plateau, the
Catskill Mountains, or the Great Dismal Swamp, The software will catch
up in time, and in the meantime I'll try to be a good neighbour and
not break things; I can experiment on my own database with my own
toolchain. But some large features are unavoidable: I'm not giving up
Lake Champlain or the Adirondack Park just because of their immense
size.

I understand that relations with a vast number of members are also
problematic, which is why I introduced a further level of breakdown
into sections on the not-quite-finished project to map
https://www.openstreetmap.org/relation/919642 .  I try to be a good
citizen with large objects, but there are large definite objects in
the field, and a rule like "no bigger than a day's walk" is going to
leave us with an urban-only map.



More information about the Tagging mailing list