Re: [Tagging] Relation for place archipelago with members place island

[Martin Koppenhoefer]

sent from a phone

> On 17. Dec 2019, at 00:28, Warin <61sundow...@gmail.com> wrote:
> 
> I personally have thoroughly rejected overlaying ways.


I am also using them rather rarely, and prefer multipolygon relations because 
they are easier to edit and easier to understand in the editor. Still we’ll 
have to live with people who see it the opposite way, for example here is a 
discussion in the German forum about discouraging the use of mp relations: 
https://forum.openstreetmap.org/viewtopic.php?id=64439

Cheers Martin ___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Warin]

On 16/12/19 21:58, Martin Koppenhoefer wrote:



Am So., 15. Dez. 2019 um 04:39 Uhr schrieb Warin 
<61sundow...@gmail.com >:


I am against overlaying one way on top of another.

The wiki suggests both tags together on a single way is acceptable;

https://wiki.openstreetmap.org/wiki/Tag:place%3Disland#Islands_in_the_sea

Given the suggested alternatives, I prefer both tags together.



what do you do when the coastline is split in several parts (there may 
be various reasons for this, some have to do with admin boundaries in 
cases where the island is split into several administrative divisions 
that end at the coastline)?


Make the place=island into a multipoygon relation. It has been done and 
looks to work, as I recall where one island had boundaries for national 
parks imposed on the coastline, ion another where beachg asnd and scrub 
were imposed on the coastline.

How can I know to which objects apply the gns tags?
Look at the history as to when the gns tags were applied, if not clear 
at the time of application then ask that mapper?


Overlaying ways is one of the possibilities (personally I do not 
prefer it, but it is generally accepted, especially as the only 
alternative in many cases are relations, which also are contested), 
and you didn't provide any reasons why it should be thoroughly rejected.


I asked for ideas. I personally have thoroughly rejected overlaying 
ways. What others do is up to them.


There is an area that I know of where overlaid ways exist .. and it is a 
real mess to edit as some nodes only join to one of the ways... It was 
remotely mapped by an 'expert'. They objected to my deletions, which are 
based on existing data and a little knowledge rather than satellite 
imagery and guess work.


___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Martin Koppenhoefer]

Am So., 15. Dez. 2019 um 04:39 Uhr schrieb Warin <61sundow...@gmail.com>:

> I am against overlaying one way on top of another.
>
> The wiki suggests both tags together on a single way is acceptable;
>
> https://wiki.openstreetmap.org/wiki/Tag:place%3Disland#Islands_in_the_sea
>
> Given the suggested alternatives, I prefer both tags together.



what do you do when the coastline is split in several parts (there may be
various reasons for this, some have to do with admin boundaries in cases
where the island is split into several administrative divisions that end at
the coastline)?
How can I know to which objects apply the gns tags?

Overlaying ways is one of the possibilities (personally I do not prefer it,
but it is generally accepted, especially as the only alternative in many
cases are relations, which also are contested), and you didn't provide any
reasons why it should be thoroughly rejected.

Cheers
Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Warin]

I am against overlaying one way on top of another.

The wiki suggests both tags together on a single way is acceptable;

https://wiki.openstreetmap.org/wiki/Tag:place%3Disland#Islands_in_the_sea

Given the suggested alternatives, I prefer both tags together.


On 15/12/19 12:54, Martin Koppenhoefer wrote:


sent from a phone


On 15. Dec 2019, at 02:11, Paul Allen  wrote:

The inner is entirely within the outer, so it meets the letter of the law (but 
maybe not the spirit).
Can renderers cope with it?


basically you have just outer members then (one in the simplest case), 
renderers and other data consumers (geocoding etc) should not have problems, 
but it may be an extra effort to parse it (relation refers to way, way refers 
to nodes). Some mappers dismiss it for this reason. IMHO editing is much easier 
for these than it is for stacked ways, and when you want to have the way 
closely tied to the area like here, it also behaves generally as desired (when 
modifying the way, the area is automatically updated as well)

Cheers Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging




___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Martin Koppenhoefer]

sent from a phone

> On 15. Dec 2019, at 02:11, Paul Allen  wrote:
> 
> The inner is entirely within the outer, so it meets the letter of the law 
> (but maybe not the spirit).
> Can renderers cope with it?


basically you have just outer members then (one in the simplest case), 
renderers and other data consumers (geocoding etc) should not have problems, 
but it may be an extra effort to parse it (relation refers to way, way refers 
to nodes). Some mappers dismiss it for this reason. IMHO editing is much easier 
for these than it is for stacked ways, and when you want to have the way 
closely tied to the area like here, it also behaves generally as desired (when 
modifying the way, the area is automatically updated as well)

Cheers Martin 
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Paul Allen]

On Sun, 15 Dec 2019 at 00:32, Martin Koppenhoefer 
wrote:

 Or you could create a multipolygon relation for the island by adding the
> coastline as outer way
>

That's something I hadn't thought of.  I just experimented in iD (didn't
save the result, it was
pure invention) and it seems to work.  It certainly makes it a lot easier
to edit in iD, although
perhaps a little confusing if you encounter one of these without knowing
why it was done that
way (and some mappers, it is said, are scared of multipolygons).  Is it
legitimate, though?
The inner is entirely within the outer, so it meets the letter of the law
(but maybe not the spirit).
Can renderers cope with it?

I'm in two minds on this one.  It makes maintenance (in iD and maybe other
editors) easier.
But it's confusing if you don't know why it's been done that way.  And it
feels clunky and
bodgetastic.  But while there's a rule against tagging for the renderer
there isn't (as yet) a
rule against tagging for the editor.

I'll wait until I see what others say before I decide whether or not to use
the trick.  But I commend
you for your ingenuity, whatever they say.

-- 
Paul
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Martin Koppenhoefer]

sent from a phone

> On 15. Dec 2019, at 00:15, Warin <61sundow...@gmail.com> wrote:
> 
> There would not be too many coast lines with names, fewer again with a 
> continuous coasts that have names.
> So I would think the name applies to the island, not the coast.


you can think it is most probable, yes, it’s not modelled though. What about 
the other tags (gns)?

Cheers Martin 
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Martin Koppenhoefer]

sent from a phone

> On 15. Dec 2019, at 00:14, Paul Allen  wrote:
> 
> You could go with the one object/one way method.  Draw an area for the 
> place=island.  On top of
> that draw a closed way for the coastline.  It's extra work, and a bit 
> tedious, but not too bad. 


yes, you could either draw a second way on top (with Josm you’d only have to 
draw 2 points and with F (follow) you could do it in seconds), although the 
result IMHO is harder to edit and the overlap is harder to notice/may pass 
unnoticed. Or you could create a multipolygon relation for the island by adding 
the coastline as outer way - if you’d have to split the coastline for other 
reasons this would still perfectly work without further work in common editors  
(splitting of MP members is typically handled automatically by editing software 
with regard to the relation)


Cheers Martin 
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Warin]

On 15/12/19 09:25, Martin Koppenhoefer wrote:


Looking at the individual islands, it is kind of problematic that the 
linear coastline and the area place=island are not distinguishable: 
you can’t tell whether that’s the name of the coastline or the island 
or both, e.g. here

https://www.openstreetmap.org/way/3932#map=8/-3.222/144.098



There would not be too many coast lines with names, fewer again with a 
continuous coasts that have names.

So I would think the name applies to the island, not the coast.



___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Paul Allen]

On Sat, 14 Dec 2019 at 22:27, Martin Koppenhoefer 
wrote:

Looking at the individual islands, it is kind of problematic that the
> linear coastline and the area place=island are not distinguishable: you
> can’t tell whether that’s the name of the coastline or the island or both,
> e.g. here
> https://www.openstreetmap.org/way/3932#map=8/-3.222/144.098
>

You could go with the one object/one way method.  Draw an area for the
place=island.  On top of
that draw a closed way for the coastline.  It's extra work, and a bit
tedious, but not too bad.  However,
the problem with that is when it becomes necessary to alter one object but
not the other (such as
adding tags to make the island a nature reserve) as iD makes that harder
than it need be (I have
no idea how josm handles it as I find it a worse fit to my way of thinking
than iD).

Unless I've missed something (always possible) iD has no way of selecting
one of two or more
coincident objects: you get whichever one it decides to give you.  If
either of the objects has one
node that isn't shared with the other, then you can hunt around for a line
segment that gives you
the object you want.  If not, you have to disconnect a node, move it a
little way, so you have line
segments that aren't shared (you may have to move more than one node if
there are more than
two object, it depends how lucky you are), make your changes, then move the
node(s) back.

Actually, it's also a problem with objects that aren't entirely coincident,
if they're large, such as
forest bounding a heath..  You may have to move the map a long way to find
where they diverge
so that you can select one rather than the other.

What is obviously needed in iD is something like assigning control-click to
cycle through the
possible coincident objects.  I suggested as much many months ago, but it
was ignored.  And
that was before I seemed to have entered iD's general "ignore these people"
list for disagreeing
with them about something.

The above is somewhat off-topic here, especially as iD has effectively
added this mailing
list to its "ignore these people" list, but editor UIs influence tagging
decisions by mappers.
It may be desirable to use two objects to handle islands but if editors
make it a pain then it will
rarely happen.

-- 
Paul
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Martin Koppenhoefer]

sent from a phone

>> On 14. Dec 2019, at 12:46, Christoph Hormann  wrote:
> That looks correct, archipelagos are normal multipolygon relations.  


+1


> Building them from the same coastline ways that are used to map the 
> individual islands is the established method for mapping them.

Looking at the individual islands, it is kind of problematic that the linear 
coastline and the area place=island are not distinguishable: you can’t tell 
whether that’s the name of the coastline or the island or both, e.g. here 
https://www.openstreetmap.org/way/3932#map=8/-3.222/144.098

Or to what the the gns-tags apply...


Cheers Martin ___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Relation for place archipelago with members place island

[Christoph Hormann]

On Saturday 14 December 2019, Warin wrote:
>
> I think this is ok. But is there a better way?
>
> The particular relation is 55737 the Schouten Islands.

You mean

https://www.openstreetmap.org/relation/557367

That looks correct, archipelagos are normal multipolygon relations.  
Building them from the same coastline ways that are used to map the 
individual islands is the established method for mapping them.

-- 
Christoph Hormann
http://www.imagico.de/

___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

[Tagging] Relation for place archipelago with members place island

[Warin]

I came across a tagged relation that had no physical tags and concluded 
that it is collection of islands, an archipelago.


So I have tagged them as such.

The members are tagged as islands and coastline on the same ways.

I think this is ok. But is there a better way?

The particular relation is 55737 the Schouten Islands.


___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging