road relations

classic Classic list List threaded Threaded
10 messages Options
Reply | Threaded
Open this post in threaded view
|

road relations

Jez Nicholson
Has anyone else come across relations grouping road assets? i.e. the road itself plus shops, buildings, street objects? e.g. https://www.openstreetmap.org/relation/1866997 Has this format become accepted elsewhere in the world or is it experimental?

Regards,
              Jez

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: road relations

Andrew Hain
It is documented at https://wiki.openstreetmap.org/wiki/Relation:associatedStreet , the terracer plugin used to create it a lot but now doesn’t by default. The Germans have been stripping it out of the database recently [ https://forum.openstreetmap.org/viewtopic.php?id=65510 ] and I’d be relaxed if we did the same.

--
Andrew



From: Jez Nicholson <[hidden email]>
Sent: 01 June 2019 11:10
To: Talk-GB
Subject: [Talk-GB] road relations
 
Has anyone else come across relations grouping road assets? i.e. the road itself plus shops, buildings, street objects? e.g. https://www.openstreetmap.org/relation/1866997 Has this format become accepted elsewhere in the world or is it experimental?

Regards,
              Jez

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: road relations

Great Britain mailing list
In reply to this post by Jez Nicholson
Hi

I've yet to hear a valid reasoning for this relation type. It's much more beneficial to add addresses instead.

There appears an increasing tendency to collect almost anything together into a relation. See public-transport's 'stop_area' as another example This is not why relations were conceived. It just adds duplication, confusion & errors. 

Personally I would delete associatedStreet.

DaveF

On 01/06/2019 11:10, Jez Nicholson wrote:
Has anyone else come across relations grouping road assets? i.e. the road
itself plus shops, buildings, street objects? e.g.
https://www.openstreetmap.org/relation/1866997 Has this format become
accepted elsewhere in the world or is it experimental?

Regards,
              Jez



_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb


_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: road relations

Gareth L

I was about to say, relations of this manner seem duplicitous of simply having an address.

 

Street objects.. like bins and benches might make a bit of sense. I don’t think I’ve ever seen a street address on a bench node. But I’m fairly sure a query could be crafted to detect the nearest way to get that information, should it be required.

 

Gareth

 


From: Dave F via Talk-GB <[hidden email]>
Sent: Saturday, June 1, 2019 11:29:33 AM
To: [hidden email]
Subject: Re: [Talk-GB] road relations
 
Hi

I've yet to hear a valid reasoning for this relation type. It's much more beneficial to add addresses instead.

There appears an increasing tendency to collect almost anything together into a relation. See public-transport's 'stop_area' as another example This is not why relations were conceived. It just adds duplication, confusion & errors. 

Personally I would delete associatedStreet.

DaveF

On 01/06/2019 11:10, Jez Nicholson wrote:
Has anyone else come across relations grouping road assets? i.e. the road
itself plus shops, buildings, street objects? e.g.
https://www.openstreetmap.org/relation/1866997 Has this format become
accepted elsewhere in the world or is it experimental?

Regards,
              Jez



_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb


_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: road relations

Colin Smale
In reply to this post by Great Britain mailing list

On 2019-06-01 12:29, Dave F via Talk-GB wrote:

Hi

I've yet to hear a valid reasoning for this relation type. It's much more beneficial to add addresses instead.

There appears an increasing tendency to collect almost anything together into a relation. See public-transport's 'stop_area' as another example This is not why relations were conceived. It just adds duplication, confusion & errors.

Relations are great to represent real-world relations that cannot be inferred (reliably) from the other data in OSM. Often a geometrical relation exists, such as a node inside a polygon, but not always.

OSM loves to allow things to be inferred from the data, but there is usually a way of entering the attributes/relationships explicitly as well, for the cases where the heuristics fall down.

The wiki says about relations: "Relations are used to model logical (and usually local) or geographic relationships between objects. They are not designed to hold loosely associated but widely spread items. It would be inappropriate, for instance, to use a relation to group 'All footpaths in East Anglia'.Why don't PT stop_areas fit with this?

C.


_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: road relations

Colin Smale
In reply to this post by Gareth L

On 2019-06-01 12:34, Gareth L wrote:

I was about to say, relations of this manner seem duplicitous of simply having an address.

 Using only the street name to link objects is unreliable. A street can be divided into multiple segments. Think of a residential side-road with the same name as the road it branches from. A house on the corner may be part of (i.e. front gate leads to) the main road, or may be part of the side road. The location of the front door (entrance=main on building outline?) is also unreliable. Only the route of the front path would give you the answer.

C.


_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: road relations

Great Britain mailing list
You're talking about a different subject, which 'associatedstreets' won't resolve.

DaveF
 
On 01/06/2019 12:06, Colin Smale wrote:
On 2019-06-01 12:34, Gareth L wrote:

I was about to say, relations of this manner seem duplicitous of simply having an address.
 Using only the street name to link objects is unreliable. A street can
be divided into multiple segments. Think of a residential side-road with
the same name as the road it branches from. A house on the corner may be
part of (i.e. front gate leads to) the main road, or may be part of the
side road. The location of the front door (entrance=main on building
outline?) is also unreliable. Only the route of the front path would
give you the answer. 

C.


_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb


_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: road relations

Great Britain mailing list
In reply to this post by Colin Smale


On 01/06/2019 12:00, Colin Smale wrote:

> Relations are great to represent real-world relations that cannot be
> inferred (reliably) from the other data in OSM. Often a geometrical
> relation exists, such as a node inside a polygon, but not always.
>
> OSM loves to allow things to be inferred from the data, but there is
> usually a way of entering the attributes/relationships explicitly as
> well, for the cases where the heuristics fall down.
>
> The wiki says about relations: "Relations are used to model logical (and
> usually local) or geographic relationships between objects. They are not
> designed to hold loosely associated but widely spread items. It would be
> inappropriate, for instance, to use a relation to group 'All footpaths
> in East Anglia'.Why don't PT stop_areas fit with this?

As all items have co-ordinates, OSM is geospatially aware; and of course
any objects with the same value tags are already 'collected together' &
searchable.

'stop_areas' - Bins are irrelevant to routing from A to B. Do any
routers use them?

DaveF

_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: road relations

Neil Matthews
In reply to this post by Andrew Hain

Not a fan -- for the more prosaic issue of what happens when you split the street -- I don't think any of the editors will automatically reassign the buildings.


Neil


On 01/06/2019 11:24, Andrew Hain wrote:
It is documented at https://wiki.openstreetmap.org/wiki/Relation:associatedStreet , the terracer plugin used to create it a lot but now doesn’t by default. The Germans have been stripping it out of the database recently [ https://forum.openstreetmap.org/viewtopic.php?id=65510 ] and I’d be relaxed if we did the same.


_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
Reply | Threaded
Open this post in threaded view
|

Re: road relations

Colin Smale
In reply to this post by Great Britain mailing list

On 2019-06-01 13:32, Dave F via Talk-GB wrote:

You're talking about a different subject, which 'associatedstreets' won't resolve.

Are you sure? Maybe you would restate concisely the problem as you see it. The relation linked to in Jez' original post was type=associatedStreet and he actually referred to a broader range of objects, many of which don't have addresses as such. Why should an explicit link between a building (or any other object) and the street it is associated with be a bad thing, if computational geometry may lead to an incorrect result? I would not like to think that the accuracy of the location of these objects should be compromised so it yields the "expected" results for a given set of circumstances.

C.


_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb