Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Joachim]

2015-09-13 23:38 GMT+02:00 Paul Norman :
> On 9/10/2015 5:20 AM, Joachim wrote:
>>
>> Define on the wiki page of highway=motorway_link that oneway=* must
>> also be tagged for every motorway_link. If not tagged, the oneway=*
>> status of this way is undefined.
>
> Explicitly tagging oneway on links is preferable for obvious reasons, but
> you need to be careful with must, which is wrong for two reasons.
>
> - The wiki can document, but not set out requirements, as people can ignore
> the current state of the wiki.
> - Your next sentence discusses the lack of oneway
> - There is not a concept of formal validity, so must doesn't apply
> - Data consumers will make their own decisions

Your concerns are valid and I changed the tone of the proposal with a
rename from "obligatory oneway" to "no default oneway". I know the the
meaning of "MUST" from IETF RfCs, but "SHOULD" would be more
appropriate here.

The first sentence about the proposal is now: "Strongly recommend
explicit tagging of oneway=* on highway=motorway_link." "

I also put this sentence in:
"The goal of this proposal is removing the implied oneway=yes on
highway=motorway_link from documentation. The following implied
default oneway=no is also undesireable and could lead to dangerous
situations in navigations. "

The statement about the routing was already changed, so I will put
this on for voting soon if no other objections are coming.

http://wiki.openstreetmap.org/wiki/Proposed_features/Motorway_link_no_default_oneway

Regards, Joachim

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Joachim]

2015-09-14 2:40 GMT+02:00 Richard Welty :
> quite. there are sections of motorway_link highways along the taconic
> parkway in NY which are two way and so lack oneway tags. now it's not that 
> hard
> to go through and fix it, but i'm reasonably sure this is not the only place 
> where
> this situation exists.

Most of Taconic Parkway is trunk and trunk_link which never implied
oneway. You should go through in order to improve safety.

> if you impose a restriction like this, then routing will be broken for
> some not yet
> identified set of links for an unknown period of time. nobody is going
> to do that.

Routing is already dangerously broken on some of the ~1400
motorway_link without oneway=* in North America since Mapquest and
Graphhopper don't imply oneway. Without turn restrictions, which I
doubt existing, they might lead you the wrong way.
Example routing: http://preview.tinyurl.com/qxyavwd
Overpass: http://overpass-turbo.eu/s/bu9

I checked a good amount of the North America Overpass query and many
of them, but no majority, implied oneway=no.

Regards, Joachim

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Greg Troxel]

Richard Welty  writes:

> On 9/13/15 5:38 PM, Paul Norman wrote:
>> On 9/10/2015 5:20 AM, Joachim wrote:

>>> Tools to help enforcing the obligatory usage:
>>> [...]
>>> - No routing over undefined oneways

>> The chances of anyone implementing this in their routing engine are
>> approximately zero.

> quite. there are sections of motorway_link highways along the taconic
> parkway in NY which are two way and so lack oneway tags. now it's not
> that hard to go through and fix it, but i'm reasonably sure this is
> not the only place where this situation exists.

My impression is that for motorway_link, 2-way traffic is unusual.   So
I wonder about a maproulette challenge that tries to identify
motorway_links that
  - lack a one_way tag
  - appear to be 2-way from geometry (the hard part)


I would also think that motorway_links that are 2-way would connect to
1-way segments for the actual on/offramp; that could help in identifying
them.  If they are right angle with a stop sign, that seems like a clue
for trunk vs motorway.


pgpSWom2Ij9ZK.pgp
Description: PGP signature
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Richard Welty]

On 9/13/15 5:38 PM, Paul Norman wrote:
> On 9/10/2015 5:20 AM, Joachim wrote:
>
>> Tools to help enforcing the obligatory usage:
>> [...]
>> - No routing over undefined oneways
> The chances of anyone implementing this in their routing engine are
> approximately zero.
quite. there are sections of motorway_link highways along the taconic
parkway
in NY which are two way and so lack oneway tags. now it's not that hard
to go through
and fix it, but i'm reasonably sure this is not the only place where
this situation
exists.

if you impose a restriction like this, then routing will be broken for
some not yet
identified set of links for an unknown period of time. nobody is going
to do that.

richard

-- 
rwe...@averillpark.net
 Averill Park Networking - GIS & IT Consulting
 OpenStreetMap - PostgreSQL - Linux
 Java - Web Applications - Search




signature.asc
Description: OpenPGP digital signature
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Paul Norman]

On 9/10/2015 5:20 AM, Joachim wrote:

Define on the wiki page of highway=motorway_link that oneway=* must
also be tagged for every motorway_link. If not tagged, the oneway=*
status of this way is undefined.


Explicitly tagging oneway on links is preferable for obvious reasons, 
but you need to be careful with must, which is wrong for two reasons.


- The wiki can document, but not set out requirements, as people can 
ignore the current state of the wiki.

- Your next sentence discusses the lack of oneway
- There is not a concept of formal validity, so must doesn't apply
- Data consumers will make their own decisions


Tools to help enforcing the obligatory usage:
[...]
- No routing over undefined oneways
The chances of anyone implementing this in their routing engine are 
approximately zero.


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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Joachim]

> Though I agree in principle with the idea of making tagging more
> explicit, how big of a practical concern is this? i.e. how many times in
> the real world is motorway_link a two-way road?

This is quite common in some parts parts of Europe. Here an Overpass
Turbo link which covers south-western Germany, Switzerland and parts
of France. The are 186 motorway_link ways with oneway=no:
http://overpass-turbo.eu/s/bp5
An usual design of a motorway exit in Germany has a shared section
near the lower road and then splits up nearer to the motorway (shaped
like a "Y").

Regards, Joachim

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Joachim]

Considering that most replies where not in favour of dropping routing
over "undefined oneway" I changed the sentence about routers:
"- For routing purposes no recommendation for ways with undefined
oneway is made. A provider should decide on it's own considering the
documentation history and current data."

The main part of the proposal is about the requirement of explicit
tagging, so I'm going with the consensus about routing. If you have
better wording, feel free to change the sentence.

Regards, Joachim

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Paul Johnson]

On Fri, Sep 11, 2015 at 7:41 AM, moltonel  wrote:

> > If not tagged, the oneway=*
> >status of this way is undefined.
>
> You wont gain anything by de-defining the "oneway=no" default value.
> Consumers (routers, renderers, whatever) will not be swayed by a wiki page.
> They might look at stats and decide themselves what the absence of a oneway
> tag means, but a wiki proposal is never going to influence that decision.


I believe theres at least a couple renderers (including our own mapnik)
that assumes oneway=yes when oneway is undefined on a motorway link.
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Mateusz Konieczny]

On Fri, 11 Sep 2015 12:41:36 +
moltonel  wrote:

> Consumers (routers, renderers, whatever) will not be swayed by a wiki
> page. They might look at stats and decide themselves what the absence
> of a oneway tag means, but a wiki proposal is never going to
> influence that decision.

Documentation on wiki is one of main sources during development of map
style.

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Mateusz Konieczny]

On Thu, 10 Sep 2015 14:20:43 +0200
Joachim  wrote:

> I drafted up a proposal about oneway=* for highway=motorway_link.
> Please comment.
> http://wiki.openstreetmap.org/wiki/Proposed_features/Motorway_link_obligatory_oneway

http://wiki.openstreetmap.org/wiki/Proposed_features/Motorway_link_obligatory_oneway#Rationale
is quite convincing that explicit tagging would be a good idea.

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Shawn K. Quinn]

On Thu, 2015-09-10 at 14:20 +0200, Joachim wrote:
> I drafted up a proposal about oneway=* for highway=motorway_link.
> Please comment.
> http://wiki.openstreetmap.org/wiki/Proposed_features/Motorway_link_obligatory_oneway
> 
> Proposal:
> Define on the wiki page of highway=motorway_link that oneway=* must
> also be tagged for every motorway_link. If not tagged, the oneway=*
> status of this way is undefined.

Though I agree in principle with the idea of making tagging more
explicit, how big of a practical concern is this? i.e. how many times in
the real world is motorway_link a two-way road?

> - For rendering purposes ways with undefined oneway should be
> displayed like the default, i.e. without oneway arrows.

This I support...

> - For routing purposes it is recommended to not route over ways with
> undefined oneway since any assumption may be wrong and it would be
> best to correct the data.

This I don't. In a lot of cases the likely direction of a motorway_link
can be inferred by the angle of the junction, even if not explicitly
tagged.

> - In map editors undefined oneway should be displayed as tagging
> error.

This makes sense, but will cause a fair amount of grief while the
existing data is fixed.

-- 
Shawn K. Quinn 


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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Kieron Thwaites]

> Though I agree in principle with the idea of making tagging more
> explicit, how big of a practical concern is this? i.e. how many times in
> the real world is motorway_link a two-way road?

While I agree such a case is rare, it is possible.

See: http://www.openstreetmap.org/way/68433570 -- while this example
is a trunk_link and not a motorway_link, the trunk road in question
may be upgraded to motorway one day (and thus the trunk_link).

--K

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[moltonel]

On 10 September 2015 13:20:43 GMT+01:00, Joachim  wrote:
>Proposal:
>Define on the wiki page of highway=motorway_link that oneway=* must
>also be tagged for every motorway_link.

Sounds fair.

> If not tagged, the oneway=*
>status of this way is undefined.

You wont gain anything by de-defining the "oneway=no" default value. Consumers 
(routers, renderers, whatever) will not be swayed by a wiki page. They might 
look at stats and decide themselves what the absence of a oneway tag means, but 
a wiki proposal is never going to influence that decision.

>- For routing purposes it is recommended to not route over ways with
>undefined oneway since any assumption may be wrong and it would be
>best to correct the data.

That's a very bad idea. Routers implementing it would skip a motorway exit (and 
lenghten the trip greatly) because of a missing tag. I'd much rather get to the 
exchange, see that the router is suggesting a link i cannot take, and drive 
around to find a nearby link I can use. Still frustrating but less time wasted.

>- In map editors undefined oneway should be displayed as tagging error.

And *that* is the actually usefull thing to do, instead of the proposal. File a 
bug to the major editors and QA tools suggesting to flag motorway links without 
a oneway tag as an warning.

Even better: if the software has the means to do it, flag that warning for any 
object having tag foo but not tag bar, if 95% of foo tags in the db are 
accompanied by a bar tag.
-- 
Vincent Dp

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Paul Johnson]

On Fri, Sep 11, 2015 at 7:07 AM, Kieron Thwaites 
wrote:

> > Though I agree in principle with the idea of making tagging more
> > explicit, how big of a practical concern is this? i.e. how many times in
> > the real world is motorway_link a two-way road?
>
> While I agree such a case is rare, it is possible.
>
> See: http://www.openstreetmap.org/way/68433570 -- while this example
> is a trunk_link and not a motorway_link, the trunk road in question
> may be upgraded to motorway one day (and thus the trunk_link).


Annoyingly, there's confusion in the US on whether that would be called a
motorway right now already because of official classifications and a few
backwater DOTs that don't differentiate between partially controlled and
fully controlled in calling something a freeway...
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Paul Johnson]

On Fri, Sep 11, 2015 at 6:56 AM, Shawn K. Quinn 
wrote:

> On Thu, 2015-09-10 at 14:20 +0200, Joachim wrote:
> > I drafted up a proposal about oneway=* for highway=motorway_link.
> > Please comment.
> >
> http://wiki.openstreetmap.org/wiki/Proposed_features/Motorway_link_obligatory_oneway
> >
> > Proposal:
> > Define on the wiki page of highway=motorway_link that oneway=* must
> > also be tagged for every motorway_link. If not tagged, the oneway=*
> > status of this way is undefined.
>
> Though I agree in principle with the idea of making tagging more
> explicit, how big of a practical concern is this? i.e. how many times in
> the real world is motorway_link a two-way road?


It's enough of an edge case I'm in favor of there not being a reasonable
assumption.  Most are one-way, there's a considerable number that aren't.
Immediately coming to mind are multiple ramps joining US:OK:Turnpike to
surface highways, particularly when not within a major urban center (such a
configuration allows a single toll taker to provide information, receipts,
change and collect tolls from a single booth where mainline toll collection
is not feasible).

Ideally checking motorway_link for oneway=* and warning for values other
than yes or no would be be handled by validators (something I also support
for motorway and motorway_link for foot=* and bicycle=* since there's a
very wide mix of both in North America, making blanket assumptions
generally bad on this continent).

> - For routing purposes it is recommended to not route over ways with
> > undefined oneway since any assumption may be wrong and it would be
> > best to correct the data.
>
> This I don't. In a lot of cases the likely direction of a motorway_link
> can be inferred by the angle of the junction, even if not explicitly
> tagged.


Agreed.


> > - In map editors undefined oneway should be displayed as tagging
> > error.
>
> This makes sense, but will cause a fair amount of grief while the
> existing data is fixed.
>

This is a band aid that needs to be ripped off, and sooner is better than
later.  Just look at the fun we're having killing route refs on ways (as
opposed to route=road relations) dinosaur...
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Mateusz Konieczny]

On Fri, 11 Sep 2015 08:40:21 -0500
Paul Johnson  wrote:

> On Fri, Sep 11, 2015 at 7:41 AM, moltonel  wrote:
> 
> > > If not tagged, the oneway=*
> > >status of this way is undefined.
> >
> > You wont gain anything by de-defining the "oneway=no" default value.
> > Consumers (routers, renderers, whatever) will not be swayed by a
> > wiki page. They might look at stats and decide themselves what the
> > absence of a oneway tag means, but a wiki proposal is never going
> > to influence that decision.
> 
> 
> I believe theres at least a couple renderers (including our own
> mapnik) that assumes oneway=yes when oneway is undefined on a
> motorway link.

openstreetmap-carto is not assuming oneway on highway=motorway_link
https://github.com/gravitystorm/openstreetmap-carto/issues/1820

Also, please avoid using name "mapnik". Mapnik is an open source toolkit
for rendering maps - not one map style.

"openstreetmap-carto" or "default map style" or "standard map style" is
a better name.

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[Richard Fairhurst]

Mateusz Konieczny wrote:
> Documentation on wiki is one of main sources during development of 
> map style.

You mean of the openstreetmap-carto style, which is just one of many.

Richard




--
View this message in context: 
http://gis.19327.n5.nabble.com/New-proposal-Obligatory-tagging-of-oneway-on-motorway-link-tp5854428p5854502.html
Sent from the Tagging mailing list archive at Nabble.com.

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

Re: [Tagging] New proposal: Obligatory tagging of oneway on motorway_link

[moltonel 3x Combo]

On 11/09/2015, Mateusz Konieczny  wrote:
> On Fri, 11 Sep 2015 12:41:36 +
> moltonel  wrote:
>
>> Consumers (routers, renderers, whatever) will not be swayed by a wiki
>> page. They might look at stats and decide themselves what the absence
>> of a oneway tag means, but a wiki proposal is never going to
>> influence that decision.
>
> Documentation on wiki is one of main sources during development of map
> style.

It is indeed an important source for some consumers, but for something
like this stats are much more important than the wiki. You can go
ahead and work on that proposal, I just don't think it's an efficient
way to improve the situation.

And since you're talking about map styles, this proposal explicitly
says that motoway_link without a oneway tag should render the same as
oneway=no. In other words, this proposal doesn't change the status quo
for map styles, it doesn't concern them.

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