Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Andrew Guertin]

On 10/14/2015 05:21 PM, Minh Nguyen wrote:

I've been guilty of mistakenly joining state boundaries to the Ohio River's
thalweg in the past, and by now I've had to correct those boundaries on
several occasions. It's unfortunate that few mappers are aware of these
complexities. The full situation is spelled out in a wiki page:


I'd suggest also tagging the ways involved in these boundaries with
  note=
to provide another way for overzealous editors to catch themselves. 
Although given the effort involved I'd probably do this only on ways 
there's reason to edit anyway.


--Andrew

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Minh Nguyen]

Mike Thompson  gmail.com> writes:

> 
> Sometimes, such as in the case of the boundary between the US states of
Ohio and Kentucky, it is the low water mark on one bank[1] (in this case the
court held that it was the low water mark of the north bank of the Ohio
River in 1792, not the present low water mark of the north bank, and
therefore the boundary and the river should not share geometry in this case).

I've been guilty of mistakenly joining state boundaries to the Ohio River's
thalweg in the past, and by now I've had to correct those boundaries on
several occasions. It's unfortunate that few mappers are aware of these
complexities. The full situation is spelled out in a wiki page:

http://wiki.openstreetmap.org/wiki/Ohio_River

If any other natural features tend to attract misinformed edits, I suggest
writing up a similar page so you can easily point mappers to it in changeset
comments etc.

Ironically, I was recently involved in a minor car collision onboard a ferry
crossing this river. The police officer who responded had to call around to
verify that he had jurisdiction. Perhaps if his department had distributed
OSM-based maps, we could've been on our way a bit sooner. ;-)

-- 
m...@nguyen.cincinnati.oh.us


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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Andrew Guertin]

On 10/14/2015 03:49 AM, Badita Florin wrote:

Our task is to delete all the existing admin_level=6 boundaries and start
fresh


Since this doesn't seem to have been discussed either here or on the 
imports list before*, how confident are you that the new data is better 
than the current data in OSM?



* I looked at the thread "Mexico's Administrative Divisions Import 
Project 1"



--Andrew

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Andrew Guertin]

On 10/14/2015 03:49 AM, Badita Florin wrote:

This way is a highway and at the same time is part of the relation of a
boundary. This seems invalid since it merges two types of features on the
same way instead of keeping a logical separation between two different
things. Is this a valid way? What if the highway is modified ? since the
highway is not a legal boundary and just happens to overlap the real
boundary, so if the highway  is changed for any reason, it will modify the
boundary along with it. So what's the valid thing to do here? Duplicate the
way to save the highway way and keep a way for the boundary separated?,


I won't get in to the best way to accomplish this technically, but I 
suggest you remove the existing boundary information in whatever way 
works for you while leaving non-boundary information intact, and then 
upload your new information keeping it separate from any other 
(non-boundary) objects.


As other people have discussed, it can be very hard (requiring legal 
research or even court decisions) to know whether a boundary IS a 
certain feature and will change if the feature changes or merely 
currently follows a feature and will stay put if the feature changes. 
Uploading the boundaries as separate objects is not wrong and provides 
the vast majority of the value. If anyone is motivated to do the legal 
research and connect things when appropriate, they can do that later.


--Andrew

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Andrew Guertin]

On 10/14/2015 04:05 AM, Sebastiaan Couwenberg wrote:

On 14-10-15 09:49, Badita Florin wrote:

Our task is to delete all the existing admin_level=6 boundaries and start
fresh, but this seems much more things needs to happen before you do this.


Don't delete the existing boundaries, update them to match the new
reality using the ReplaceGeometry feature in JOSM for example. When the
data for shared nodes is available, it will disconnect the other ways
from the boundary way being replaced leaving the other ways as they were.


I disagree with this suggestion, and I think the original plan of 
deleting the existing ways or tags and uploading new ones is better. 
Reasons:


1) The value of using Replace Geometry is very low for this case. The 
reason for doing so would be to make life easier for anyone who wants to 
know what OSM previously thought the boundaries are. Very few people 
will want to know that, especially since it won't provide any context 
for understanding the new, imported data. And for those few that ever 
will, the tools still exist and work fine.


2) Replace Geometry won't work well. To provide a meaningful consistency 
of history, there needs to be a roughly one-to-one correspondence 
between new objects and old.


To explain this with an example: Imagine the county boundaries are 
currently mapped, with one way between each pair of counties, and a 
relation for each county collecting the appropriate ways. Now add 
detail, mapping out the boundaries for each town. Each county relation 
is now formed by a larger number of smaller ways which are the town 
boundaries. What should happen to the original ways that were used for 
county boundaries? They don't correspond to anything in the new scheme, 
so there's nothing for Replace Geometry to do that makes sense.


--Andrew

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Mike Thompson]

On Wed, Oct 14, 2015 at 4:27 AM, Colin Smale  wrote:

>
> The boundary is where the government says it is...
>
Correct and there is a difference between "delimiting" (marking on a map or
specifying coordinates) and demarcation (placing or referencing physical
features on the ground - e.g. survey markers or monumentation).   I can't
find a citation for this, but I recall from my studies that where the two
disagree, demarcation takes precedence, even when it conflicts with treaty.
This is great for OSM, because the demarcation should be verifiable on the
ground (and sometimes from overhead imagery too). Therefore, importing
boundaries from a government source may be a good start, the best method is
to  survey the situation.

Mike

>
>
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Mike Thompson]

On Wed, Oct 14, 2015 at 3:43 AM, Colin Smale  wrote:

>
>
>
> A boundary couldn't be "the river" as a river has non-zero width. It might
> be the "centre line", "deepest line", "fastest flowing bit" . but it
> cannot be "the river" without further qualification.
>
Sometimes when a river legally forms a boundary it is the thalweg (deepest
part) that is referenced.

Sometimes, such as in the case of the boundary between the US states of
Ohio and Kentucky, it is the low water mark on one bank[1] (in this case
the court held that it was the low water mark of the north bank of the Ohio
River in 1792, not the present low water mark of the north bank, and
therefore the boundary and the river should not share geometry in this
case).

Mike

[1] https://supreme.justia.com/cases/federal/us/444/335/case.html
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Martin Koppenhoefer]

2015-10-14 14:40 GMT+02:00 Christoph Hormann :

> The general convention in OSM, also for boundaries, is to map the actual
> situation on the ground, that is which areas are actually administred
> by which authority.
>


this often doesn't help though, because in remote areas there is nothing to
"administer". Some time ago the case of the Mont Blanc came up on a local
mailing list, more precisely in which language the name should be tagged
(the decision was to put both names into the name tag:
http://www.openstreetmap.org/node/281399025/history although France is
claiming to "administer" the peak alone).

Cheers,
Martin
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Christoph Hormann]

On Wednesday 14 October 2015, Colin Smale wrote:
> > A large fraction of 'authorative' sources of boundary data have
> > very little to do with the legal/contractual definition of the
> > boundary.  I would probably go as far as saying the most inaccurate
> > boundaries in OSM come from authorative sources.
>
> I would be interested in some supporting evidence for this...

OK - i need to constrict that - the most inaccurate boundaries in OSM 
are maritime boundaries which do not generally come from authorative 
sources.  The most inaccurate land boundaries are probably from 
non-authorative sources like CIA database and LSIB.

Without knowing the actual demarcation of a boundary this is generally 
difficult to assess and there are no boundaries imported from 
authorative sources around here but things like

http://www.openstreetmap.org/way/191851483

are likely inaccurate since nodes are placed away from the feature the 
boundary seems to follow and are located at positions that would not be 
well suited for demarcation.

> >> The boundary is where the government says it is...
> >
> > Not in OSM - see the 'on the ground rule'.  For OSM the boundary is
> > what locals treat as the boundary.
>
> That's a different boundary then. The area between the two might be
> "disputed", [...]

No, disputed is when people on one side of the boundary have a different 
idea of where the boundary is than those on the other side.

The general convention in OSM, also for boundaries, is to map the actual 
situation on the ground, that is which areas are actually administred 
by which authority.

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

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Jóhannes Birgir Jensson]

It's not that simple. I work in an government agency and the issue of 
boundaries rises often, both for public and private issues.


Iceland is an advanced nation regarding technology adaption. However 
boundaries are not all clearly defined as GIS vectors and many of them 
are disputed.


There are many natural reserve areas, some small and some large - yet we 
do not have definite and GPS accurate definitions of them available yet. 
Most of them are based upon text descriptions of areas that were defined 
previously for another purpose. Travelling the country and documenting 
place names has proven to be an interesting experience - a survey was 
done earlier this century.


A common scenario is the definition of where two estates are adjacent to 
each other. The legal document defines the boundary as lying between 
place names #1 and #2, from where a direct line through to #3 and 
followed by the middle of a river. Not a single GPS point in that. Then 
we go to find where the place names are, it should be easy right? Not 
really, the new residents are unsure of where place name #1 is, it could 
possibly be that tiny hillock amongst many, or it could be the one 3 
hillocks over, next to the flat rock. So we already are unsure of our 
starting point, and each point has similar issues. Sometimes the place 
is defined as 50 paces from another place and then the measurement 
sticks used originally have been found and found wanting or being larger 
than they should...


A flash flood from a glacier melting changes the river regularly, 
shifting it faster around than in more stable geographical areas. And 
look at that, a volcano has just created a new lava field, which changed 
the course of a river by the virtue of completely closing off its 
previous path (Wikipedia: Nornahraun and Holuhraun and the river Jökulsá 
á Fjöllum).


There have been court battles and disputes all over the country and 
municipalities also have disputes, some based on these textual 
descriptions of places no one knows where are or where the features have 
shifted several times over since the original line was drawn in the 
unknown place.


So the government will tell you that the boundary is between these place 
names, and isn't always able to tell you if these place names are at the 
same spot as they originally were. There is work ongoing of converting 
these into GPS co-ordinates but that could have to go through courts in 
some cases.


So the government doesn't always have the definite answer via GPS 
points.




Þann 14.10.2015 11:43, Colin Smale reit:

Well, although it is definitely not unknown, I think it probably is
fair to call it rare in the grand scheme of things... The vast
majority of administrative boundaries in the world are not disputed,
and the ones that are, are more likely to be the national borders
(admin_level=2) than internal provincial or municipal boundaries, of
which there are many, many more.

On 2015-10-14 13:35, Martin Koppenhoefer wrote:


sent from a phone


Am 14.10.2015 um 12:27 schrieb Colin Smale
:

The boundary is where the government says it is...


yes, but the governments of adjoining states having different ideas
about this is also not rare.

Cheers
Martin


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



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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Colin Smale]

 

Well, although it is definitely not unknown, I think it probably is fair
to call it rare in the grand scheme of things... The vast majority of
administrative boundaries in the world are not disputed, and the ones
that are, are more likely to be the national borders (admin_level=2)
than internal provincial or municipal boundaries, of which there are
many, many more. 

On 2015-10-14 13:35, Martin Koppenhoefer wrote: 

> sent from a phone
> 
>> Am 14.10.2015 um 12:27 schrieb Colin Smale :
>> 
>> The boundary is where the government says it is...
> 
> yes, but the governments of adjoining states  having different ideas about 
> this is also not rare.
> 
> Cheers 
> Martin
 ___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Martin Koppenhoefer]

sent from a phone

> Am 14.10.2015 um 12:27 schrieb Colin Smale :
> 
> The boundary is where the government says it is...


yes, but the governments of adjoining states  having different ideas about this 
is also not rare.

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Colin Smale]

 

On 2015-10-14 13:04, Christoph Hormann wrote: 

> On Wednesday 14 October 2015, Colin Smale wrote: 
> 
>> Boundaries are often downloadable from authoritative sources. The
>> downloadable data is however not always the legal definition of the
>> boundary, but derived from that definition [...]
> 
> A large fraction of 'authorative' sources of boundary data have very 
> little to do with the legal/contractual definition of the boundary.  I 
> would probably go as far as saying the most inaccurate boundaries in 
> OSM come from authorative sources.

I would be interested in some supporting evidence for this... 

>> The boundary is where the government says it is...
> 
> Not in OSM - see the 'on the ground rule'.  For OSM the boundary is what 
> locals treat as the boundary.

That's a different boundary then. The area between the two might be
"disputed", or there might be a difference between "de jure" and "de
facto" boundaries - which are both right, just in different contexts.
There is room in OSM for both perspectives. 

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Christoph Hormann]

On Wednesday 14 October 2015, Colin Smale wrote:
> Boundaries are often downloadable from authoritative sources. The
> downloadable data is however not always the legal definition of the
> boundary, but derived from that definition [...]

A large fraction of 'authorative' sources of boundary data have very 
little to do with the legal/contractual definition of the boundary.  I 
would probably go as far as saying the most inaccurate boundaries in 
OSM come from authorative sources.

>
> The boundary is where the government says it is...

Not in OSM - see the 'on the ground rule'.  For OSM the boundary is what 
locals treat as the boundary.

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

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Paul Norman]

On 10/14/2015 1:23 AM, Frederik Ramm wrote:

You'd have to research how the boundary is defined. If there is some
sort of legal definition that goes "the boundary has the following
geometry: from lat/lon A to lat/lon B to lat/lon C...", independent of
the river or highway, then it makes sense to have two different
geometries. But if the legal definition goes "the municipality of X
extends until the middle of the river Y" then it would be wrong to have
two different geometries in OSM.


There's another possibility, found more commonly in remote areas, and 
not often in Europe: none of the above.


Boundaries are not always rigorously defined, and it may not be set if 
the boundary precisely follows the river or not, or in the case of 
multiple similar branches, which branch is followed.


There's also the case where the boundary is defined to be the river, but 
not follow when the river shifts, or only follow some types of river shifts.


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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Colin Smale]

 

Boundaries are often downloadable from authoritative sources. The
downloadable data is however not always the legal definition of the
boundary, but derived from that definition - either by surveying if the
definition is descriptive, or by generalisation as the full level of
detail is too much for the download (for whatever reason), or by
reprojection (if the boundary is legally defined in a different datum
such as OSGB36). But the result of all that is a set of coordinates
which we effectively cannot dispute. 

The boundary is where the government says it is... 

//colin 

On 2015-10-14 12:17, Christoph Hormann wrote: 

> On Wednesday 14 October 2015, Maarten Deen wrote: 
> Academic detail. "Is" the boundary the river, or is the boundary a
> thing
> its own right, the geometry of which is described by the river? I
> think you can argue either way. 
> And the legal part can be different too. It can be that the boundary
> is the river and it will change when the river changes, it can also
> be that the boundary has been defined as the river at a point in time
> and if the river changes after that point, the boundary does not
> change with it.

Note practically this is usually 'academic detail' as well - most 
demarcated boundaries are not represented by actual demarcation points 
in OSM but by some approximately drawn line.  In case of boundaries at 
rivers this rarely gets worse when you attach the boundary to the 
accurately mapped river.  Examples:

http://www.openstreetmap.org/#map=15/32.0077/35.5299
http://www.openstreetmap.org/#map=14/0.7489/29.9702 ___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Christoph Hormann]

On Wednesday 14 October 2015, Maarten Deen wrote:
> >
> > Academic detail. "Is" the boundary the river, or is the boundary a
> > thing
> > its own right, the geometry of which is described by the river? I
> > think you can argue either way.
>
> And the legal part can be different too. It can be that the boundary
> is the river and it will change when the river changes, it can also
> be that the boundary has been defined as the river at a point in time
> and if the river changes after that point, the boundary does not
> change with it.

Note practically this is usually 'academic detail' as well - most 
demarcated boundaries are not represented by actual demarcation points 
in OSM but by some approximately drawn line.  In case of boundaries at 
rivers this rarely gets worse when you attach the boundary to the 
accurately mapped river.  Examples:

http://www.openstreetmap.org/#map=15/32.0077/35.5299
http://www.openstreetmap.org/#map=14/0.7489/29.9702

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

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Colin Smale]

 

A boundary couldn't be "the river" as a river has non-zero width. It
might be the "centre line", "deepest line", "fastest flowing bit" .
but it cannot be "the river" without further qualification. 

On 2015-10-14 11:31, Frederik Ramm wrote: 

> Hi,
> 
> On 10/14/2015 10:56 AM, Martin Koppenhoefer wrote: 
> 
>> (If this happens - if the boundary is defined by the river or the
>> highway - then you still have various options of modeling this, for
>> example having two ways share the same nodes, 
>> 
>> wouldn't this kind of modelling be "wrong"? If the boundary IS the
>> river, there shouldn't be 2 different objects there, should they?
> 
> Academic detail. "Is" the boundary the river, or is the boundary a thing
> its own right, the geometry of which is described by the river? I think
> you can argue either way.
> 
> Bye
> Frederik
 ___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Maarten Deen]

On 2015-10-14 11:31, Frederik Ramm wrote:

Hi,

On 10/14/2015 10:56 AM, Martin Koppenhoefer wrote:

(If this happens - if the boundary is defined by the river or the
highway - then you still have various options of modeling this, 
for

example having two ways share the same nodes,

wouldn't this kind of modelling be "wrong"? If the boundary IS the
river, there shouldn't be 2 different objects there, should they?


Academic detail. "Is" the boundary the river, or is the boundary a 
thing

its own right, the geometry of which is described by the river? I think
you can argue either way.


And the legal part can be different too. It can be that the boundary is 
the river and it will change when the river changes, it can also be that 
the boundary has been defined as the river at a point in time and if the 
river changes after that point, the boundary does not change with it.


And since it is easy to change ways in OSM and people will change 
visible features, either based on own visual confirmation or based on 
aerial photo's (which as we all know may have an offset) and will 
probably not be so eager to change "official" boundaries, I would alwas 
separate the two.


Regards,
Maarten


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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Frederik Ramm]

Hi,

On 10/14/2015 10:56 AM, Martin Koppenhoefer wrote:
> (If this happens - if the boundary is defined by the river or the
> highway - then you still have various options of modeling this, for
> example having two ways share the same nodes, 
> 
> wouldn't this kind of modelling be "wrong"? If the boundary IS the
> river, there shouldn't be 2 different objects there, should they?

Academic detail. "Is" the boundary the river, or is the boundary a thing
its own right, the geometry of which is described by the river? I think
you can argue either way.

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frede...@remote.org  ##  N49°00'09" E008°23'33"

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Martin Koppenhoefer]

2015-10-14 10:23 GMT+02:00 Frederik Ramm :

> (If this happens - if the boundary is defined by the river or the
> highway - then you still have various options of modeling this, for
> example having two ways share the same nodes,
>



wouldn't this kind of modelling be "wrong"? If the boundary IS the river,
there shouldn't be 2 different objects there, should they?

Cheers,
Martin
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Frederik Ramm]

Hi,

On 10/14/2015 09:49 AM, Badita Florin wrote:
> This way is a highway and at the same time is part of the relation of a
> boundary. This seems invalid since it merges two types of features on
> the same way instead of keeping a logical separation between two
> different things. Is this a valid way? What if the highway is modified ?
> since the highway is not a legal boundary and just happens to overlap
> the real boundary, so if the highway  is changed for any reason, it will
> modify the boundary along with it.

This is not automatically wrong. It is possible that boundaries are
actually defined by highways or other features (rivers, for example).

> So what's the valid thing to do here?

You'd have to research how the boundary is defined. If there is some
sort of legal definition that goes "the boundary has the following
geometry: from lat/lon A to lat/lon B to lat/lon C...", independent of
the river or highway, then it makes sense to have two different
geometries. But if the legal definition goes "the municipality of X
extends until the middle of the river Y" then it would be wrong to have
two different geometries in OSM.

(If this happens - if the boundary is defined by the river or the
highway - then you still have various options of modeling this, for
example having two ways share the same nodes, or putting the
river/highway into the boundary relation. The latter seems more commonly
used but both are valid.)

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frede...@remote.org  ##  N49°00'09" E008°23'33"

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Martin Koppenhoefer]

2015-10-14 9:49 GMT+02:00 Badita Florin :

> This way is a highway and at the same time is part of the relation of a
> boundary. This seems invalid since it merges two types of features on the
> same way instead of keeping a logical separation between two different
> things. Is this a valid way? What if the highway is modified ? since the
> highway is not a legal boundary and just happens to overlap the real
> boundary, so if the highway  is changed for any reason, it will modify the
> boundary along with it.



This really depends on the definition of the boundary. If the highway IS
legally the boundary, the boundary might also change when the highway
changes (more likely for natural features maybe, like rivers, peaks or
coastlines). If instead the legal boundary is defined separately (e.g. by
coordinates or poles on the ground) and "just happens" to coincide with the
highway position then we should model 2 distinct features in OSM (and a
modification of the highway should not modify the boundary as well). You
cannot assume that the highway isn't the legal boundary unless you find the
actual legal definition for it and can verify the situation. Both
alternatives exist in the real world.

Cheers,
Martin
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Colin Smale]

 

How is the boundary legally defined? If it is a set of coordinates or a
line on a map, then there is no intrinsic link with the line of the
highway. If the highway is realigned, this will not (automatically)
affect the boundary. This may have already happened in the past, so the
lines are "almost" colinear but not quite. 

However, if the boundary is legally defined in a more descriptive
manner, such as "following the centre line of highway X" then there is a
real link between the line of the highway and the line of the boundary,
and one cannot be changed without changing the other. 

Where boundaries appear to follow a highway, they are not always aligned
to the centre line. This can give headaches for road maintenance, police
jurisdiction etc. The boundary may be following the edge of the
carriageway, or a hedge/ditch a couple of metres away, so the whole road
is actually in one administrative area. 

So all in all, I would suggest it would be better to not share ways, and
not share nodes either - unless you are very sure that the boundary and
the highway are legally linked together. 

--colin 

On 2015-10-14 09:49, Badita Florin wrote: 

> It is interesting the things that you discover when trying to do the import 
> of a whole county, in this case, Mexico 
> 
> Our task is to delete all the existing admin_level=6 boundaries and start 
> fresh, but this seems much more things needs to happen before you do this.  
> 
> There are over 500 nodes or ways that share a common node or follow the same 
> path with the old boundaries that we want to remove, so we can add the new 
> boundaries. One interesting topic had have sprung up is this, what are the 
> procedures for dealing with boundaries that run along a road, waterway. 
> 
> A query that we had build to detect this kind of shared nodes and ways looks 
> like this http://overpass-turbo.eu/s/bZA 
> 
> You can adapt the script and share your results, to see if there are other 
> examples that can be included here 
> 
> Now it has come up a question, what happens for example with cases like this. 
> 
> A way that is two things simultaneously, for example a highway and a boundary 
> at the same time. There are some cases like these in the state of Michoacan 
> and I can't just de-glue the bad nodes because the way  itself seems to be 
> two things at the same time.  This is the issue in question, your query 
> identified the following two nodes: 
> 
> Nodes 1856092007 [1] and 1856092002 [2] , which limit the following  way [3] 
> between such nodes. This way is a highway and at the same time is part of the 
> relation of a boundary. This seems invalid since it merges two types of 
> features on the same way instead of keeping a logical separation between two 
> different things. Is this a valid way? What if the highway is modified ? 
> since the highway is not a legal boundary and just happens to overlap the 
> real boundary, so if the highway  is changed for any reason, it will modify 
> the boundary along with it. So what's the valid thing to do here? Duplicate 
> the way to save the highway way and keep a way for the boundary separated?, 
> I've found similar questions [4] by other users and they indicate it isn't 
> valid but I need a more official argument because the user is upset if we 
> remove this kind of ways from relations 
> ᐧ 
> 
> ___
> talk mailing list
> talk@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
 

Links:
--
[1]
http://t.sidekickopen24.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XYg4XrmfHMQByL0f6nqlW3LyBkH56dL4Zf8p2wPs02?t=http%3A%2F%2Fwww.openstreetmap.org%2Fnode%2F1856092007&si=6117088740507648&pi=d9ea1099-68c4-44a7-9853-3b1c4a86d057
[2]
http://t.sidekickopen24.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XYg4XrmfHMQByL0f6nqlW3LyBkH56dL4Zf8p2wPs02?t=http%3A%2F%2Fwww.openstreetmap.org%2Fnode%2F1856092002&si=6117088740507648&pi=d9ea1099-68c4-44a7-9853-3b1c4a86d057
[3]
http://t.sidekickopen24.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XYg4XrmfHMQByL0f6nqlW3LyBkH56dL4Zf8p2wPs02?t=http%3A%2F%2Fwww.openstreetmap.org%2Fway%2F174970090&si=6117088740507648&pi=d9ea1099-68c4-44a7-9853-3b1c4a86d057
[4]
http://t.sidekickopen24.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XYg4XrmfHMQByL0f6nqlW3LyBkH56dL4Zf8p2wPs02?t=https%3A%2F%2Fhelp.openstreetmap.org%2Fquestions%2F7563%2Fwaterway-as-administrative-boundary-shared-way&si=6117088740507648&pi=d9ea1099-68c4-44a7-9853-3b1c4a86d057___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Sebastiaan Couwenberg]

On 14-10-15 09:49, Badita Florin wrote:
> Our task is to delete all the existing admin_level=6 boundaries and start
> fresh, but this seems much more things needs to happen before you do this.

Don't delete the existing boundaries, update them to match the new
reality using the ReplaceGeometry feature in JOSM for example. When the
data for shared nodes is available, it will disconnect the other ways
from the boundary way being replaced leaving the other ways as they were.

> This way is a highway and at the same time is part of the relation of a
> boundary. This seems invalid since it merges two types of features on the
> same way instead of keeping a logical separation between two different
> things. Is this a valid way? What if the highway is modified ? since the
> highway is not a legal boundary and just happens to overlap the real
> boundary, so if the highway  is changed for any reason, it will modify the
> boundary along with it. So what's the valid thing to do here? Duplicate the
> way to save the highway way and keep a way for the boundary separated?,

In the Americas people seem to be fond of using natural features as
boundary ways, in The Netherlands we keep these strictly separate. The
administrative boundary may follow a river for example, but these are
two distinct features that happen to share a similar geometry. At the
highest zoom level in JOSM you'll see that the river way doesn't share
the nodes from the boundary and if they do that's an issue to be fixed
by disconnecting the shared node(s), each feature has its own way and
nodes. Because natural features tend to change when the administrative
boundaries do not, we have to keep them separate.

Kind Regards,

Bas

-- 
 GPG Key ID: 4096R/6750F10AE88D4AF1
Fingerprint: 8182 DE41 7056 408D 6146  50D1 6750 F10A E88D 4AF1

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

Re: [OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Maarten Deen]

On 2015-10-14 09:49, Badita Florin wrote:


Nodes 1856092007 [1] and 1856092002 [2] , which limit the following
way [3] between such nodes. This way is a highway and at the same time
is part of the relation of a boundary. This seems invalid since it
merges two types of features on the same way instead of keeping a
logical separation between two different things. Is this a valid way?


I have seen this more often. Is it a valid way of mapping? Sure, why 
not. Is it prudent? I don't think so, precisely because of your 
concerns:



What if the highway is modified ? since the highway is not a legal
boundary and just happens to overlap the real boundary, so if the
highway  is changed for any reason, it will modify the boundary along
with it.


That is why it you have to be very cautious in connecting different 
kinds of objects on the same nodes. I've seen this also on multiple 
occasions with landuses and roads. Not only does this make editting a 
bit awkward (more difficult to select the object you want to edit), it 
also is unclear what the meaning of it is.



So what's the valid thing to do here? Duplicate the way to
save the highway way and keep a way for the boundary separated?, I've
found similar questions [4] by other users and they indicate it isn't
valid but I need a more official argument because the user is upset if
we remove this kind of ways from relations


I assume you are not doing automated edits? Then I would just remove the 
boundary tags from the road and remove the way from the boundary 
relation and draw/import the boundary new. To have two nodes on exactly 
the same spot is also not very nice. The JOSM validator will give a 
warning about that and then you risk that people are going to merge 
them.


Regards,
Maarten


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

[OSM-talk] open question about boundaries sharing nodes with ways or nodes

[Badita Florin]

It is interesting the things that you discover when trying to do the import
of a whole county, in this case, Mexico

Our task is to delete all the existing admin_level=6 boundaries and start
fresh, but this seems much more things needs to happen before you do this.

There are over 500 nodes or ways that share a common node or follow the
same path with the old boundaries that we want to remove, so we can add the
new boundaries. One interesting topic had have sprung up is this, what are
the procedures for dealing with boundaries that run along a road, waterway.

A query that we had build to detect this kind of shared nodes and ways
looks like this http://overpass-turbo.eu/s/bZA

You can adapt the script and share your results, to see if there are other
examples that can be included here

Now it has come up a question, what happens for example with cases like
this.

A way that is two things simultaneously, for example a highway and a
boundary at the same time. There are some cases like these in the state of
Michoacan and I can't just de-glue the bad nodes because the way  itself
seems to be two things at the same time.  This is the issue in question,
your query identified the following two nodes:

Nodes 1856092007

and 1856092002

, which limit the following  way

between such nodes.
This way is a highway and at the same time is part of the relation of a
boundary. This seems invalid since it merges two types of features on the
same way instead of keeping a logical separation between two different
things. Is this a valid way? What if the highway is modified ? since the
highway is not a legal boundary and just happens to overlap the real
boundary, so if the highway  is changed for any reason, it will modify the
boundary along with it. So what's the valid thing to do here? Duplicate the
way to save the highway way and keep a way for the boundary separated?,
I've found similar questions

by other users and they indicate it isn't valid but I need a more official
argument because the user is upset if we remove this kind of ways from
relations
ᐧ
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk