[Tagging] Using multipolygons to map bays in Alaska

Daniel Koć daniel at xn--ko-wla.pl
Sun Nov 18 04:42:11 UTC 2018


W dniu 17.11.2018 o 12:27, Frederik Ramm pisze:

> But I felt in this situation, they had overstepped their mandate,
> *especially* because they were not reacting to something that people
> were doing, but actively creating a new feature ("hey, you can now have
> huge named bays") and at the same time adding the data to OSM to
> illustrate their new feature.


The reality is more complex, so I'd like to show the other part of this
story.

During one of my map reviews around the world I have encountered Port
Phillip Bay (near Melbourne) not being rendered. This is quite old bay
relation - it has 3 years now:

https://www.openstreetmap.org/relation/1221199

I was interested how this water area is named and it looked bizarre for
me that this information is completely invisible on default map. Using
areas for bays was documented on the wiki, the numbers were OK and still
raising, the patch was soon accepted, so it all looked OK for me.


> And let's be totally clear here: A nice label on the sea is all that
> Daniel wanted. He's not a maritime scientist who for some reason needs
> the exact extent of Bothninan Bay - he went through the time-consuming
> exercise of combining more than 1600 coastline pieces into one huge
> polygon which is difficult to handle for data processors and editors
> alike JUST TO PLACE A LABEL.


Nope, having a sane label was just a part of my motivation. It was also
"eating my own dog food", which I do with some features just to see how
does it really work.

When I saw a random dot somewhere inside big water body as a result of
asking Nominatim instead of a general shape I expected, I was disgusted
and frustrated. It didn't make sense for me to shrink a 2D place with
size of a country into 0-dimensional object, placed there without any
clear reason and without a hope for any sane data operation other than
name searching. I was also happy that this place will be much closer to
reality after this fix - you don't need to be a scientist to appreciate
that.


> It is only a matter of time until they start labelling natural=sea
> polygons and people then create relations with 100,000 members for the
> Atlantic.


I really hope one day we will have some conventions for bigger water
objects. Limited testing on something smaller and more solid like bays
was also one of my goals. Currently OpenStreetMap seems to be completely
unprepared for _any_ large scale mapping (other than countries and some
natural areas), so I doubt it will be anytime soon.

This includes not only geometry problems (sorry - a node for entire Asia
looks like a joke!), but also lack of convention for tagging biggest
rivers, lack of idea how to tag and render generic names of biggest
areas (3 languages in Belgium as name=* is quite strange, but which and
how many languages should we display for Atlantic?), lack of convention
for mountain ranges, geographical regions and so on...

But since we've made conventions for such artificial objects like road
types, it's possible that one day we will have some global (sic!)
agreement. IMO it's like 5 years perspective if everything goes smooth
(which I don't think is realistic vision, unfortunately, ~10 years is
much more probable).


> (1) the OSM-Carto project and Daniel have overstepped their mandate as
> the maintainers of our style, and should have sought a wider consensus
> on this before acting;


For me a project as big and distributed as OSM is rather ecosystem than
a single project. Any action can have unexpected consequences, but lack
of action will also have serious consequences. Constant discussing
everything and care taken for every small detail almost killed OSM Carto
in 2015. It took years to make more people contribute again. I hope this
new team will find good balance and will upset people less than currently.


> (2) the decision they have made is not a good solution for the
> cartographic problem they wanted to solve;


Nobody is forced by documentation to choose this or the other solution,
so now we have a chance to see how people will use these tools. At least
they have a real choice now and I believe this will help OSM to grow in
a more organic way.


> (3) the decision they have made will lead to people creating huge
> polygons that will often break, make coastline editing harder, and have
> at least one totally made-up edge.


Almost every water body connected to another (river/sea, lake/river,
bay/ocean, and even stream/river) has at least one unverifiable border,
because - well... - you can't see the ground truth on the water; unless
there's a dam for example. We just ignore it for small water bodies and
the whole coastlines. However it does not seem to me as a sustainable
solution, we need some convention.

Every country or national park border is a fragile large beast, which
may be only approximately visible on the ground. You might see some
boundary markers, but mostly there is no continuous line and we are
connecting the dots at best or using some documentation from
authorities, which is a mediocre solution verifiability-wise; it's not
possible to draw them using GPS device or aerial imagery, like with
roads for example.

But there's a reason why we keep them as areas rather than nodes - they
are better approximated by virtual lines than single dots and it's more
useful for people to treat them like that.


> (4) Frederik has been an utter dick to try and start the discussion by
> deleting the Bothany Bay polygon, instead of simply raising it here. It
> was wrong, I'm sorry.


Thanks for a funny comment, it offloads some tension in a heated debate.
:-)

I was really upset than, but now I'm happy that I didn't act with a
knee-jerk reaction. It was also good thing that I had no time to discuss
it with you on the changeset page - it's much better here, on the list,
with multiple points of view.


-- 
"Excuse me, I have some growing up to do" [P. Gabriel]





More information about the Tagging mailing list