site relations for city walls?

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
38 messages Options
12
Reply | Threaded
Open this post in threaded view
|

site relations for city walls?

dieterdreist
Someone has made a site relation for the Aurelian citywalls in Rome.
Does this make sense to you?
We‘re speaking of a generally linear object of many kilometers length, in parts fragmented / interrupted.

Cheers Martin

sent from a phone
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

yo paseopor
Big sense, nerver forget.
What about that?


Health (more now than never) and maps
yopaseopor

On Sun, Jul 12, 2020 at 8:44 PM Martin Koppenhoefer <[hidden email]> wrote:
Someone has made a site relation for the Aurelian citywalls in Rome.
Does this make sense to you?
We‘re speaking of a generally linear object of many kilometers length, in parts fragmented / interrupted.

Cheers Martin

sent from a phone
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

dieterdreist


sent from a phone

On 12. Jul 2020, at 20:53, yo paseopor <[hidden email]> wrote:



it’s a type=collection though, not a site. And questionable in parts, as the Berlin wall is often/sometimes still visible from fragments or remaining infrastructure (also small things like lights, light poles, cable attachments, etc.), but in many places there are no visible traces left...

In Rome the situation is different because the walls mostly have survived, and where arterial roads have cut through they are often holes in the wall, (sometimes also removed entirely though).

Cheers Martin 

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

Taskar Center
In reply to this post by yo paseopor

Why is the relation type on the Berlin Wall a “collection” rather than “boundary”?

Thanks,
Anat


Sent from my mobile. Please excuse brevity and typos.

On Jul 12, 2020, at 11:52 AM, yo paseopor <[hidden email]> wrote:

Big sense, nerver forget.
What about that?


Health (more now than never) and maps
yopaseopor

On Sun, Jul 12, 2020 at 8:44 PM Martin Koppenhoefer <[hidden email]> wrote:
Someone has made a site relation for the Aurelian citywalls in Rome.
Does this make sense to you?
We‘re speaking of a generally linear object of many kilometers length, in parts fragmented / interrupted.

Cheers Martin

sent from a phone
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

dieterdreist


sent from a phone

> On 12. Jul 2020, at 21:14, Taskar Center <[hidden email]> wrote:
>
> Why is the relation type on the Berlin Wall a “collection” rather than “boundary”?


it’s a collection of remaining traces of a boundary (which btw was never a „line“ in the geometric sense, because there always was a buffer that made it a zone around the boundary, and it never was the actual political boundary, which was towards the Western Berlin side, because the wall was built inside the Eastern Berlin area).
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

Andy Townsend
In reply to this post by Taskar Center
On 12/07/2020 20:13, Taskar Center wrote:
>
> Why is the relation type on the Berlin Wall a “collection” rather than
> “boundary”?

Over its history as an object in OSM it's had a whole variety of tags. 
Different people have said "This is important!  We should render it!"
and have sometimes tried to do that by adjusting the tags until they
found something that rendered.  Other people have tried to change tags
to better reflect the current reality.
http://osm.mapki.com/history/relation.php?id=6651797 tells the story.

Personally, I don't think that "boundary" would be a good relation type
for it because it isn't one - it's the route of a former barrier within
a boundary.  Compare
https://www.openstreetmap.org/relation/6651797#map=19/52.51616/13.37698 
with the suburb boundary just to the west.

Best Regards,

Andy



_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

voschix
I do consider a site relation a fitting approach for a city wall.

On Sun, 12 Jul 2020, 22:35 Andy Townsend, <[hidden email]> wrote:
On 12/07/2020 20:13, Taskar Center wrote:
>
> Why is the relation type on the Berlin Wall a “collection” rather than
> “boundary”?

Over its history as an object in OSM it's had a whole variety of tags. 
Different people have said "This is important!  We should render it!"
and have sometimes tried to do that by adjusting the tags until they
found something that rendered.  Other people have tried to change tags
to better reflect the current reality.
http://osm.mapki.com/history/relation.php?id=6651797 tells the story.

Personally, I don't think that "boundary" would be a good relation type
for it because it isn't one - it's the route of a former barrier within
a boundary.  Compare
https://www.openstreetmap.org/relation/6651797#map=19/52.51616/13.37698
with the suburb boundary just to the west.

Best Regards,

Andy



_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

dieterdreist


sent from a phone

On 13. Jul 2020, at 00:11, Volker Schmidt <[hidden email]> wrote:

I do consider a site relation a fitting approach for a city wall.


its use would also go against the wiki definition which states: „ This relation is not to be used in cases where the element can be represented by one or more areas and neither linear ways nor nodes outside these areas would have to be included or excluded from within these areas“

clearly the remains of the Aurelian walls can be nicely  represented by areas. Indeed it seems a good representation to map them as buildings, and people including myself have started to do it some time ago.

Generally I believe the requirement for a site relation that its constituting parts should be in the same town, is not strict enough. A handful of objects scattered around in a town are not a „site“. A site means things are concentrated around a point, and when there are more things in the other side of the town that somehow belonged to it, they would be considered off site, i.e. their relationship would come from other aspects, not because they are part of the same „site“.

Cheers Martin 

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

Yves-3
@Martin, the quote from the wiki really looks like a multipolygon definition. Would those walls be mapped as a multipolygon instead?

Why do you say "A site means things are concentrated around a point", sites relation helps to map disjoint elements, but I don't think I saw anything about their repartition. Also it certainly makes no sense to have sites extending over extremely large areas.
Yves

Le 13 juillet 2020 01:14:40 GMT+02:00, Martin Koppenhoefer <[hidden email]> a écrit :


sent from a phone

On 13. Jul 2020, at 00:11, Volker Schmidt <[hidden email]> wrote:

I do consider a site relation a fitting approach for a city wall.


its use would also go against the wiki definition which states: „ This relation is not to be used in cases where the element can be represented by one or more areas and neither linear ways nor nodes outside these areas would have to be included or excluded from within these areas“

clearly the remains of the Aurelian walls can be nicely  represented by areas. Indeed it seems a good representation to map them as buildings, and people including myself have started to do it some time ago.

Generally I believe the requirement for a site relation that its constituting parts should be in the same town, is not strict enough. A handful of objects scattered around in a town are not a „site“. A site means things are concentrated around a point, and when there are more things in the other side of the town that somehow belonged to it, they would be considered off site, i.e. their relationship would come from other aspects, not because they are part of the same „site“.

Cheers Martin 

--
Envoyé de mon appareil Android avec Courriel K-9 Mail. Veuillez excuser ma brièveté.
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

dieterdreist


sent from a phone

> On 13. Jul 2020, at 05:59, Yves <[hidden email]> wrote:
>
> Why do you say "A site means things are concentrated around a point", sites relation helps to map disjoint elements, but I don't think I saw anything about their repartition.


it is my interpretation of the term “site” and a kind of proposal to modify the site relation description in the wiki


> Also it certainly makes no sense to have sites extending over extremely large areas.


at least not if they aren’t “compact”

Cheers Martin
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

Lionel Giard
The aurelian walls left today are not one continuous element, so the site relation is practical as it allows to group all the linear or area features that are forming the old city walls.
I used many time in cities when few ruins of the same original object exist at different places (often walls), so you can clearly link them together with this relation. ^_^

Best Regards,
Lionel

Le lun. 13 juil. 2020 à 09:00, Martin Koppenhoefer <[hidden email]> a écrit :


sent from a phone

> On 13. Jul 2020, at 05:59, Yves <[hidden email]> wrote:
>
> Why do you say "A site means things are concentrated around a point", sites relation helps to map disjoint elements, but I don't think I saw anything about their repartition.


it is my interpretation of the term “site” and a kind of proposal to modify the site relation description in the wiki


> Also it certainly makes no sense to have sites extending over extremely large areas.


at least not if they aren’t “compact”

Cheers Martin
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

voschix
In reply to this post by dieterdreist
Looking back at the history of the site relation, it looks as if the concept originated from schools, colleges, universities, airports,  and military bases for situations where the objects are not within a well defined perimeter. Documented uses include historical sites, even though they are not quoted in recent versions of the wiki page.

Reading the wiki versions, I would say the "site" relation is extremely vaguely defined.

I would think we are free to make it something useful.

At the risk of repeating myself, I believe there is a need  for something like the site relation for a whole array of more or less widely scattered objects that belong together. Just a few:
  • non-campus universities, research institutions, schools
  • the offices of public institutions (city, regional, country governments, the European Union institutions)
  • archaeological sites (aqueducts, Hadrian's Wall, the Limes in Germany, the Great Wall of China, The Iron Curtain, city walls, former Railways, former canals and other waterways, former underground mines, ...)
  • power plants (hydro-electric, wind power, ...)
  • active mines
  • distributed museums
Volker





Virus-free. www.avast.com

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

Lionel Giard
I also saw it used for parking lot that are completely separated (like on two sides of a big highway) but still part of the "same" parking technically (like the example of mall parking in different parts separated by highways). To add to the two area mapped as amenity=parking, there was a site relation grouping them (site=parking).  ^_^ 

Le lun. 13 juil. 2020 à 19:04, Volker Schmidt <[hidden email]> a écrit :
Looking back at the history of the site relation, it looks as if the concept originated from schools, colleges, universities, airports,  and military bases for situations where the objects are not within a well defined perimeter. Documented uses include historical sites, even though they are not quoted in recent versions of the wiki page.

Reading the wiki versions, I would say the "site" relation is extremely vaguely defined.

I would think we are free to make it something useful.

At the risk of repeating myself, I believe there is a need  for something like the site relation for a whole array of more or less widely scattered objects that belong together. Just a few:
  • non-campus universities, research institutions, schools
  • the offices of public institutions (city, regional, country governments, the European Union institutions)
  • archaeological sites (aqueducts, Hadrian's Wall, the Limes in Germany, the Great Wall of China, The Iron Curtain, city walls, former Railways, former canals and other waterways, former underground mines, ...)
  • power plants (hydro-electric, wind power, ...)
  • active mines
  • distributed museums
Volker





Virus-free. www.avast.com
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

Peter Elderson
Wouldn't a multipolygon with just two outers solve that parking case?
Best Peter Elderson


Op ma 13 jul. 2020 om 21:02 schreef Lionel Giard <[hidden email]>:
I also saw it used for parking lot that are completely separated (like on two sides of a big highway) but still part of the "same" parking technically (like the example of mall parking in different parts separated by highways). To add to the two area mapped as amenity=parking, there was a site relation grouping them (site=parking).  ^_^ 

Le lun. 13 juil. 2020 à 19:04, Volker Schmidt <[hidden email]> a écrit :
Looking back at the history of the site relation, it looks as if the concept originated from schools, colleges, universities, airports,  and military bases for situations where the objects are not within a well defined perimeter. Documented uses include historical sites, even though they are not quoted in recent versions of the wiki page.

Reading the wiki versions, I would say the "site" relation is extremely vaguely defined.

I would think we are free to make it something useful.

At the risk of repeating myself, I believe there is a need  for something like the site relation for a whole array of more or less widely scattered objects that belong together. Just a few:
  • non-campus universities, research institutions, schools
  • the offices of public institutions (city, regional, country governments, the European Union institutions)
  • archaeological sites (aqueducts, Hadrian's Wall, the Limes in Germany, the Great Wall of China, The Iron Curtain, city walls, former Railways, former canals and other waterways, former underground mines, ...)
  • power plants (hydro-electric, wind power, ...)
  • active mines
  • distributed museums
Volker





Virus-free. www.avast.com
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

dieterdreist
In reply to this post by voschix


sent from a phone

On 13. Jul 2020, at 19:04, Volker Schmidt <[hidden email]> wrote:

Reading the wiki versions, I would say the "site" relation is extremely vaguely defined.

I would think we are free to make it something useful.


I agree 



At the risk of repeating myself, I believe there is a need  for something like the site relation for a whole array of more or less widely scattered objects that belong together. Just a few:
  • non-campus universities, research institutions, schools
  • the offices of public institutions (city, regional, country governments, the European Union institutions)
  • archaeological sites (aqueducts, Hadrian's Wall, the Limes in Germany, the Great Wall of China, The Iron Curtain, city walls, former Railways, former canals and other waterways, former underground mines, ...)
  • power plants (hydro-electric, wind power, ...)
  • active mines
  • distributed museums


actually all of these could be „grouped“ with tags alone, e.g distributed museums could have an identifying „network“ tag (or sth similar).
For power plants a site might be appropriate, if an area does not do it and you don’t want to rely on only tags.
In theory objects like the Great Wall in China can and should be modeled as areas, although they seem to be linear in nature, they are also thick enough to „require“ an area representation in order to be well mapped in the scale of OpenStreetMap (you can walk on it). In practice we would also want a way to have preliminary mapping as a line, and mixed geometry relations. A multipolygon relation for all parts of the great wall would likely be broken every day, and would be over the member limits for relations.

Would those that are in favour of using a site relation for a linear, circular, interrupted structure, 19km long and some meters wide, also see it as a good relation type for the Chinese Great Wall?

Cheers Martin 

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

voschix


On Mon, 13 Jul 2020 at 22:56, Martin Koppenhoefer <[hidden email]> wrote:
actually all of these could be „grouped“ with tags alone, e.g distributed museums could have an identifying „network“ tag (or sth similar).
But why invent a new network tag, if we have  a site relation, waiting to be used. (I was thinking of open air museums, where the various exhibits are spread over the landscape)
For power plants a site might be appropriate, if an area does not do it and you don’t want to rely on only tags.
If you have ever looked at the complexities of a hydro-power-plant with dams, lakes, pipes, turbines deep in the mountains or in dedicated buildings . they are really complex, and only parts of it are visible on the surface.
In theory objects like the Great Wall in China can and should be modeled as areas, although they seem to be linear in nature, they are also thick enough to „require“ an area representation in order to be well mapped in the scale of OpenStreetMap (you can walk on it).
That's not true - you can walk on parts of it, other parts are completely missing, others are heaps of stones.
In practice we would also want a way to have preliminary mapping as a line, and mixed geometry relations. A multipolygon relation for all parts of the great wall would likely be broken every day, and would be over the member limits for relations.
It's not a multipolygon - it is bits and pieces, some connected, same not. Some may be linear (in first approximation).
 
Would those that are in favour of using a site relation for a linear, circular, interrupted structure, 19km long and some meters wide, also see it as a good relation type for the Chinese Great Wall?
You lost me with your question here.

Volker

Virus-free. www.avast.com

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

Lionel Giard
Wouldn't a multipolygon with just two outers solve that parking case?
Best Peter Elderson

That's a bit of a stretch of the multipolygon definition as there is no inner ring.  I never used multipolygon for anything else than complex geometry (with inner ring(s)) and that seems to be what the feature is for.

As we already have the site relation for grouping features that are part of the same thing, but disjoint, i think that it is good to use it. It also solves the problem when mappers use multipolygon for two polygons sharing the same edge (it is forming an invalid geometry), while with site relation it is not a problem. Another advantage is that it is quite easy to edit. You just need to add or remove a feature : no specific roles (yet) or order needed.

Le lun. 13 juil. 2020 à 23:29, Volker Schmidt <[hidden email]> a écrit :


On Mon, 13 Jul 2020 at 22:56, Martin Koppenhoefer <[hidden email]> wrote:
actually all of these could be „grouped“ with tags alone, e.g distributed museums could have an identifying „network“ tag (or sth similar).
But why invent a new network tag, if we have  a site relation, waiting to be used. (I was thinking of open air museums, where the various exhibits are spread over the landscape)
For power plants a site might be appropriate, if an area does not do it and you don’t want to rely on only tags.
If you have ever looked at the complexities of a hydro-power-plant with dams, lakes, pipes, turbines deep in the mountains or in dedicated buildings . they are really complex, and only parts of it are visible on the surface.
In theory objects like the Great Wall in China can and should be modeled as areas, although they seem to be linear in nature, they are also thick enough to „require“ an area representation in order to be well mapped in the scale of OpenStreetMap (you can walk on it).
That's not true - you can walk on parts of it, other parts are completely missing, others are heaps of stones.
In practice we would also want a way to have preliminary mapping as a line, and mixed geometry relations. A multipolygon relation for all parts of the great wall would likely be broken every day, and would be over the member limits for relations.
It's not a multipolygon - it is bits and pieces, some connected, same not. Some may be linear (in first approximation).
 
Would those that are in favour of using a site relation for a linear, circular, interrupted structure, 19km long and some meters wide, also see it as a good relation type for the Chinese Great Wall?
You lost me with your question here.

Volker

Virus-free. www.avast.com
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

dieterdreist


sent from a phone

> On 14. Jul 2020, at 16:55, Lionel Giard <[hidden email]> wrote:
>
> That's a bit of a stretch of the multipolygon definition as there is no inner ring.


sorry? The minimum requisite for a multipolygon is one outer ring. There must not be inner rings or multiple outer rings. It is required for any area that is made up of anything that is not just a single closed way.



>   I never used multipolygon for anything else than complex geometry (with inner ring(s)) and that seems to be what the feature is for.


If you can use areas to represent something, you should prefer them over the site relation, it is stated in bold on the site page.


Cheers Martin


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

Peter Elderson
In reply to this post by Lionel Giard
Just two outers is a regular use of multipolygon. 
If the tags of two areas are the same, you can represent two or more distinct areas as a multipolygon

If you have one area as a multipolygon with an inner, a separate closed way can be used as an extra outer, it will then get the attributes of the multipolygon.

Major renderers support this. 

One parking lot on two sides of a road is perfect for this method.

Best, Peter Elderson


Op di 14 jul. 2020 om 16:55 schreef Lionel Giard <[hidden email]>:
Wouldn't a multipolygon with just two outers solve that parking case?
Best Peter Elderson

That's a bit of a stretch of the multipolygon definition as there is no inner ring.  I never used multipolygon for anything else than complex geometry (with inner ring(s)) and that seems to be what the feature is for.

As we already have the site relation for grouping features that are part of the same thing, but disjoint, i think that it is good to use it. It also solves the problem when mappers use multipolygon for two polygons sharing the same edge (it is forming an invalid geometry), while with site relation it is not a problem. Another advantage is that it is quite easy to edit. You just need to add or remove a feature : no specific roles (yet) or order needed.

Le lun. 13 juil. 2020 à 23:29, Volker Schmidt <[hidden email]> a écrit :


On Mon, 13 Jul 2020 at 22:56, Martin Koppenhoefer <[hidden email]> wrote:
actually all of these could be „grouped“ with tags alone, e.g distributed museums could have an identifying „network“ tag (or sth similar).
But why invent a new network tag, if we have  a site relation, waiting to be used. (I was thinking of open air museums, where the various exhibits are spread over the landscape)
For power plants a site might be appropriate, if an area does not do it and you don’t want to rely on only tags.
If you have ever looked at the complexities of a hydro-power-plant with dams, lakes, pipes, turbines deep in the mountains or in dedicated buildings . they are really complex, and only parts of it are visible on the surface.
In theory objects like the Great Wall in China can and should be modeled as areas, although they seem to be linear in nature, they are also thick enough to „require“ an area representation in order to be well mapped in the scale of OpenStreetMap (you can walk on it).
That's not true - you can walk on parts of it, other parts are completely missing, others are heaps of stones.
In practice we would also want a way to have preliminary mapping as a line, and mixed geometry relations. A multipolygon relation for all parts of the great wall would likely be broken every day, and would be over the member limits for relations.
It's not a multipolygon - it is bits and pieces, some connected, same not. Some may be linear (in first approximation).
 
Would those that are in favour of using a site relation for a linear, circular, interrupted structure, 19km long and some meters wide, also see it as a good relation type for the Chinese Great Wall?
You lost me with your question here.

Volker

Virus-free. www.avast.com
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: site relations for city walls?

Peter Elderson
https://wiki.openstreetmap.org/wiki/Multipolygon_Examples  example 1.7  describes disjunct outers. 

Too bad you have to wrestle through some very complicated examples to get there if you start at the beginning. And, these complex examples should not be followed, because they advocate tying landuse to ways, borders to ways and other stuff you really should not do if you want to keep the map unbroken.

Best, Peter Elderson


Op di 14 jul. 2020 om 18:05 schreef Peter Elderson <[hidden email]>:
Just two outers is a regular use of multipolygon. 
If the tags of two areas are the same, you can represent two or more distinct areas as a multipolygon

If you have one area as a multipolygon with an inner, a separate closed way can be used as an extra outer, it will then get the attributes of the multipolygon.

Major renderers support this. 

One parking lot on two sides of a road is perfect for this method.

Best, Peter Elderson


Op di 14 jul. 2020 om 16:55 schreef Lionel Giard <[hidden email]>:
Wouldn't a multipolygon with just two outers solve that parking case?
Best Peter Elderson

That's a bit of a stretch of the multipolygon definition as there is no inner ring.  I never used multipolygon for anything else than complex geometry (with inner ring(s)) and that seems to be what the feature is for.

As we already have the site relation for grouping features that are part of the same thing, but disjoint, i think that it is good to use it. It also solves the problem when mappers use multipolygon for two polygons sharing the same edge (it is forming an invalid geometry), while with site relation it is not a problem. Another advantage is that it is quite easy to edit. You just need to add or remove a feature : no specific roles (yet) or order needed.

Le lun. 13 juil. 2020 à 23:29, Volker Schmidt <[hidden email]> a écrit :


On Mon, 13 Jul 2020 at 22:56, Martin Koppenhoefer <[hidden email]> wrote:
actually all of these could be „grouped“ with tags alone, e.g distributed museums could have an identifying „network“ tag (or sth similar).
But why invent a new network tag, if we have  a site relation, waiting to be used. (I was thinking of open air museums, where the various exhibits are spread over the landscape)
For power plants a site might be appropriate, if an area does not do it and you don’t want to rely on only tags.
If you have ever looked at the complexities of a hydro-power-plant with dams, lakes, pipes, turbines deep in the mountains or in dedicated buildings . they are really complex, and only parts of it are visible on the surface.
In theory objects like the Great Wall in China can and should be modeled as areas, although they seem to be linear in nature, they are also thick enough to „require“ an area representation in order to be well mapped in the scale of OpenStreetMap (you can walk on it).
That's not true - you can walk on parts of it, other parts are completely missing, others are heaps of stones.
In practice we would also want a way to have preliminary mapping as a line, and mixed geometry relations. A multipolygon relation for all parts of the great wall would likely be broken every day, and would be over the member limits for relations.
It's not a multipolygon - it is bits and pieces, some connected, same not. Some may be linear (in first approximation).
 
Would those that are in favour of using a site relation for a linear, circular, interrupted structure, 19km long and some meters wide, also see it as a good relation type for the Chinese Great Wall?
You lost me with your question here.

Volker

Virus-free. www.avast.com
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
12