Re: [Tagging] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-18 Thread Mateusz Konieczny via Tagging



Oct 18, 2020, 10:17 by dieterdre...@gmail.com:

>
>
> sent from a phone
>
>> On 16. Oct 2020, at 09:32, Richard Fairhurst  wrote:
>>
>> generally bicycle=dismount should be used instead, reserving bicycle=no for 
>> those circumstances where even pushing a bike is not legal (e.g. most public 
>> footpaths in England & Wales).
>>
>
>
> most bicycle=no tags out there actually mean that you cannot ride a bike, not 
> that you cannot have a bike in your pocket or be pushing a bike or carrying a 
> bike in a box or on your shoulders.
>
> I would suggest a different tag than bicycle=no for places where you cannot 
> bring a bicycle, because otherwise you will never know which interpretation 
> of bicycle=no was used by the mapper.
>
+1

At this point bicycle=no means "no cycling allowed" and trying to change meaning
would be quite hopeless.

You would need a special tag to mark which interpretation is used and resurvey 
all
bicycle=no cases. And at that point it is easier to have a new tag for rare "no 
bicycle at all
in addition to forbidding cycling"

At that point it is easier to simply invent a new tag for "no bicycle pushing".

(bicycle_pushed=no, bicycle_pushing=no and bicycle_possession=no were proposed)
 And I think at every point in OSM history, as bicycle=dismount was a duplicate 
of bicycle=no

> The wiki is unsure about the exact meaning, the bicycle=* page says it is 
> about restrictions for bicycles while the access page (older) says it is 
> about restrictions for cyclists. IMHO the most common interpretation is 
> legality of cycling/riding a bicycle.
>
""

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


Re: [Tagging] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-18 Thread Martin Koppenhoefer


sent from a phone

> On 16. Oct 2020, at 09:32, Richard Fairhurst  wrote:
> 
> generally bicycle=dismount should be used instead, reserving bicycle=no for 
> those circumstances where even pushing a bike is not legal (e.g. most public 
> footpaths in England & Wales).


most bicycle=no tags out there actually mean that you cannot ride a bike, not 
that you cannot have a bike in your pocket or be pushing a bike or carrying a 
bike in a box or on your shoulders.

I would suggest a different tag than bicycle=no for places where you cannot 
bring a bicycle, because otherwise you will never know which interpretation of 
bicycle=no was used by the mapper.

The wiki is unsure about the exact meaning, the bicycle=* page says it is about 
restrictions for bicycles while the access page (older) says it is about 
restrictions for cyclists. IMHO the most common interpretation is legality of 
cycling/riding a bicycle.

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


Re: [Tagging] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-18 Thread Emvee via Tagging

On 16/10/2020 09:06, Richard Fairhurst wrote:

It does not currently take any account of bicycle=no on a crossing,
not least because bicycle=no is a very problematic tag - generally
bicycle=dismount should be used instead, reserving bicycle=no for
those circumstances where even pushing a bike is not legal (e.g. most
public footpaths in England & Wales).


Good to hear what cycle.travel does regarding with bicycle=no/dismount
on a crossing, that is ignore it. Also good to hear your perspective on
bicycle=dismount versus bicycle=no. It makes sense but I was not aware.

On bicycle=no/dismount on highway=crossing: In >95% of the cases
bicycle=no/dismount is useless because the access rights on the
connecting ways suffice. My educated guess is that there are about 3000
crossings marked with bicycle=no/dismount while allowed to cycle over
them using the road.

brouter does take bicycle=no/dismount in node context into account, see
https://github.com/abrensch/brouter/issues/265 and that gives a moderate
penalty as the assumption is made on can unmount crossing the crossing
node.

The problem with ignoring is that "bicycle=no/dismount" on
highway=crossing is that it becomes use useless tag. The main data
consumers of openstreetmap data are map makers (who do not care) and
routers as for others the data on crossings is not complete enough for
other use. So the routing perspective is an important one.

Should routers (keep) ignoring bicycle=no/dismount on a highway=crossing
node?
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging


Re: [Tagging] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-18 Thread Mateusz Konieczny via Tagging



Oct 18, 2020, 09:44 by tagging@openstreetmap.org:

> On 15/10/2020 17:36, Jmapb via Tagging wrote:
>
>> I've always been surprised to see a footway=crossing/cycleway=crossing
>> way with the intersection node tagged as highway=crossing. There's
>> only a single physical crossing, so this seems contra to the
>> one-feature-one-element rule.
>>
>> A highway=crossing node makes sense in an area without mapped
>> footways/cycleways. But if the crossing ways are mapped, routing
>> software will need to examine the intersection node and scan the
>> properties of all highways intersecting there. It seems to make
>> tagging the node itself redundant.
>>
>> Are there really routers that require the node be tagged as well?
>>
> Routers do not need highway=crossing to detect crossings, every node
> where 3 or more ways connect is a crossing.
>
Note that there are multiple cases where 3 or more ways connect without
it being a crossing.

Even after "any case where 3 or more ways connect, with at least one being
a road and at least 2 being a footway/path/cycleway/footway" amendment
not all of them will be crossing (footway joining terminal node of road,
two footways joinining road on one side etc).

Crossing may be in situation where just one footway line joins road.

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


Re: [Tagging] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-18 Thread Martin Koppenhoefer


sent from a phone

> On 18. Oct 2020, at 09:46, Emvee via Tagging  
> wrote:
> 
> Routers do not need highway=crossing to detect crossings, every node
> where 3 or more ways connect is a crossing.


sure, but many highway=crossings are tagged on nodes where only 2 ways (or one 
going through) connect.

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


Re: [Tagging] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-18 Thread Emvee via Tagging

On 15/10/2020 17:36, Jmapb via Tagging wrote:

I've always been surprised to see a footway=crossing/cycleway=crossing
way with the intersection node tagged as highway=crossing. There's
only a single physical crossing, so this seems contra to the
one-feature-one-element rule.

A highway=crossing node makes sense in an area without mapped
footways/cycleways. But if the crossing ways are mapped, routing
software will need to examine the intersection node and scan the
properties of all highways intersecting there. It seems to make
tagging the node itself redundant.

Are there really routers that require the node be tagged as well?


Routers do not need highway=crossing to detect crossings, every node
where 3 or more ways connect is a crossing.

I also do not add highway=crossing unless I also add crossing=*


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


Re: [Tagging] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-17 Thread Adam Franco
As someone who renders a driving-focused map
 of [the most twisty] roadways, I
specifically have done exactly what Volker describes (looking at
highway=crossing nodes only).

To provide an example, my renderer walks down each vehicle-legal way and
demotes the curviness weighting for a distance in each direction whenever
it encounters a highway=crossing node on that way (or nodes with
highway=stop, highway=traffic_signals, barrier=traffic_calming, etc). This
particular map doesn't care about the geometry of footways, sidewalks,
paths, or buildings, so it can look at a much reduced data-set of just
vehicle-specific highways. If highway=crossing nodes aren't available and
crossings are only indicated on intersecting ways, then I'd have to add a
preprocessing step to build a list of all nodes that are members of a
highway=crossing way and then add that to the list of nodes tagged with
highway=crossing. I guess it's not an impossible task, but it is much more
simple to just look at nodes that are also members of the
vehicle-accessible highway ways.

I know OsmAnd can be configured to alert drivers of upcoming crossings (and
stop signs), but do not know if that router works only with nodes on the
ways of the current route or also does matching on crossing ways.

On Thu, Oct 15, 2020 at 1:06 PM Volker Schmidt  wrote:

> I don't know what the routers need, to be honest.
> I have adopted the approach happily because of the frequent two-stage
> approach. First the main road is mapped with foot/bicycle crossings as
> nodes , and at a later stage someone else may add the foot/cycleway
> details  - I did not occur to me that there may be an advantage in removing
> at that stage the already existing crossing node.
> I would also naively assume, that a car-only router does not need to
> inspect any of the foot/cycleways in the map, and can use the
> highway=crossing nodes as an indication to add small delays inthe routing.
> Anyone in the router business listening in on this conversation?
>
> On Thu, 15 Oct 2020 at 17:39, Jmapb via Tagging 
> wrote:
>
>> On 10/13/2020 6:30 PM, Kevin Kenny wrote:
>>
>> On Tue, Oct 13, 2020, 17:41 Volker Schmidt  wrote:
>>
>> I changed the crossing to the way we do it in many parts of Europe, i.e.
>>> a crossing node *and* a crossing way. This was described as an option
>>> on the highway=crossing wiki page until it was changed on 07:52, 3 October
>>> 2020by user Emvee  by
>>> addng the diagram and its description.
>>> If you don't like it, please change it back - I used it in place of a
>>> longish explanation.
>>>
>>
>> Both of those are better, thanks! The routers that I use for testing seem
>> to be aware of crossings without crossing nodes, so I too often forget to
>> tag them.
>>
>> I've always been surprised to see a footway=crossing/cycleway=crossing
>> way with the intersection node tagged as highway=crossing. There's only a
>> single physical crossing, so this seems contra to the
>> one-feature-one-element rule.
>>
>> A highway=crossing node makes sense in an area without mapped
>> footways/cycleways. But if the crossing ways are mapped, routing software
>> will need to examine the intersection node and scan the properties of all
>> highways intersecting there. It seems to make tagging the node itself
>> redundant.
>>
>> Are there really routers that require the node be tagged as well?
>>
>> Jason
>> ___
>> Tagging mailing list
>> Tagging@openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/tagging
>>
> ___
> 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] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-16 Thread Richard Fairhurst
Volker Schmidt wrote:
> I don't know what the routers need, to be honest.
> Anyone in the router business listening in on this conversation?

cycle.travel will take account of highway=crossing nodes (e.g. where a cycleway 
crosses a road), and adjust its routing weight accordingly. The adjustment is 
slightly different depending on the type of crossing and the highway= value of 
each connecting way.

It does not take any particular note of =crossing ways, other than to note that 
footway=crossing means that the rider should push.

It does not currently take any account of bicycle=no on a crossing, not least 
because bicycle=no is a very problematic tag - generally bicycle=dismount 
should be used instead, reserving bicycle=no for those circumstances where even 
pushing a bike is not legal (e.g. most public footpaths in England & Wales).

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


Re: [Tagging] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-15 Thread Volker Schmidt
I don't know what the routers need, to be honest.
I have adopted the approach happily because of the frequent two-stage
approach. First the main road is mapped with foot/bicycle crossings as
nodes , and at a later stage someone else may add the foot/cycleway
details  - I did not occur to me that there may be an advantage in removing
at that stage the already existing crossing node.
I would also naively assume, that a car-only router does not need to
inspect any of the foot/cycleways in the map, and can use the
highway=crossing nodes as an indication to add small delays inthe routing.
Anyone in the router business listening in on this conversation?

On Thu, 15 Oct 2020 at 17:39, Jmapb via Tagging 
wrote:

> On 10/13/2020 6:30 PM, Kevin Kenny wrote:
>
> On Tue, Oct 13, 2020, 17:41 Volker Schmidt  wrote:
>
> I changed the crossing to the way we do it in many parts of Europe, i.e. a
>> crossing node *and* a crossing way. This was described as an option on
>> the highway=crossing wiki page until it was changed on 07:52, 3 October
>> 2020by user Emvee  by
>> addng the diagram and its description.
>> If you don't like it, please change it back - I used it in place of a
>> longish explanation.
>>
>
> Both of those are better, thanks! The routers that I use for testing seem
> to be aware of crossings without crossing nodes, so I too often forget to
> tag them.
>
> I've always been surprised to see a footway=crossing/cycleway=crossing way
> with the intersection node tagged as highway=crossing. There's only a
> single physical crossing, so this seems contra to the
> one-feature-one-element rule.
>
> A highway=crossing node makes sense in an area without mapped
> footways/cycleways. But if the crossing ways are mapped, routing software
> will need to examine the intersection node and scan the properties of all
> highways intersecting there. It seems to make tagging the node itself
> redundant.
>
> Are there really routers that require the node be tagged as well?
>
> Jason
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
>
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging


[Tagging] Crossing tagged on both way and node (was: What does bicycle=no on a node means?)

2020-10-15 Thread Jmapb via Tagging

On 10/13/2020 6:30 PM, Kevin Kenny wrote:

On Tue, Oct 13, 2020, 17:41 Volker Schmidt mailto:vosc...@gmail.com>> wrote:

I changed the crossing to the way we do it in many parts of
Europe, i.e. a crossing node _and_ a crossing way. This was
described as an option on the highway=crossing wiki page until it
was changed on 07:52, 3 October 2020by user Emvee
 by addng the
diagram and its description.
If you don't like it, please change it back - I used it in place
of a longish explanation.


Both of those are better, thanks! The routers that I use for testing
seem to be aware of crossings without crossing nodes, so I too often
forget to tag them.


I've always been surprised to see a footway=crossing/cycleway=crossing
way with the intersection node tagged as highway=crossing. There's only
a single physical crossing, so this seems contra to the
one-feature-one-element rule.

A highway=crossing node makes sense in an area without mapped
footways/cycleways. But if the crossing ways are mapped, routing
software will need to examine the intersection node and scan the
properties of all highways intersecting there. It seems to make tagging
the node itself redundant.

Are there really routers that require the node be tagged as well?

Jason

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