Re: [Talk-GB] Turn Restrictions at roundabouts

2020-11-22 Thread Andy Townsend
Following a lack of answers to questions at 
http://resultmaps.neis-one.org/osm-discussion-comments?uid=5229644 (in 
lots of cases they've responded, but have not actually answered the 
question) with a DWG hat on I've sent them a message that they have to 
read before continuing to map at 
https://www.openstreetmap.org/user_blocks/4214 .


I'd like to know the source used for the comment "You sometimes have to 
ignore signage as they are signed with convenient numbers rather than 
the real ones"at https://www.openstreetmap.org/changeset/60806661 .


I suspect the "signage to be ignored" referred to there are the large 
blue signs telling drivers "if you want to go to the M6 south, go this 
way".  However, I'd expect the information on the 100m markers to be 
more useful.  Is anyone aware of a previous changeset based on those, or 
photos on e.g. Mapillary that might help?


Best Regards,

Andy



___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-11-11 Thread Mateusz Konieczny via Talk-GB



Nov 11, 2020, 16:04 by talk-gb@openstreetmap.org:

> >After a quick look at his edits locally he has also been removing ref
> >tags from roundabouts which seems an odd thing to do.
>
> This seems perfectly reasonable to me - the roundabout is a junction of 
> various roads and I do not consider it to be part of a referenced highway.
>
> I note that the wiki indicates that the ref should be added to roundabouts to 
> allow fluid routing, but this has relatively recently been added (April 2019) 
> and I do not agree. It smacks of tagging for the renderer (in this case a 
> routing engine). It seems bizarre to specify that for naming it should not 
> use the name of a road it connects, but it should use the ref of a road that 
> connects!
>
I reworded this recommendation in
https://wiki.openstreetmap.org/w/index.php?title=Tag:junction%3Droundabout=2059754=1972469
to 
"{{Tag|ref}} and {{Tag|int_ref}} tags from those ways should be added to that 
roundabout if roundabout is also part of that routes."
(in Poland roundabout would be part of route with assigned ref, in UK situation 
may be different,
I removed part that based in on tagging for router)
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-11-11 Thread Mike Baggaley via Talk-GB
>After a quick look at his edits locally he has also been removing ref
>tags from roundabouts which seems an odd thing to do.

This seems perfectly reasonable to me - the roundabout is a junction of various 
roads and I do not consider it to be part of a referenced highway.

I note that the wiki indicates that the ref should be added to roundabouts to 
allow fluid routing, but this has relatively recently been added (April 2019) 
and I do not agree. It smacks of tagging for the renderer (in this case a 
routing engine). It seems bizarre to specify that for naming it should not use 
the name of a road it connects, but it should use the ref of a road that 
connects!

Regards,
Mike


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-11-10 Thread Philip Barnes
On Tue, 2020-11-10 at 11:31 +, Andy Townsend wrote:
> On 10/11/2020 10:55, Jon Pennycook wrote:
> > Returning to this subject, but not necessarily at roundabouts -
> > turn 
> > restrictions are still being added even where they don't exist 
> > (apparently) - e.g. 
> > https://www.openstreetmap.org/changeset/93759133 
> >  and 
> > https://www.openstreetmap.org/changeset/93750062 
> > 
> 
> I've commented on the first of these with a DWG hat on.
> 
> The contributor here adding turn restrictions that allegedly don't
> exist 
> appears to be from https://www.sabre-roads.org.uk/ . They've been
> active 
> across several areas of England; it might be worth locals taking a
> look 
> at some of the others to check that they match on-the-ground reality.
> 
After a quick look at his edits locally he has also been removing ref
tags from roundabouts which seems an odd thing to do.

He has also made this change https://www.osm.org/changeset/90024395

This does not follow the on the ground rule, this section is signed
M42.

Previously discussed and fixed by Andy after a previous sabre mapper
mapper made a similar edit https://www.osm.org/changeset/60806661.

Phil (trigpoint


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-11-10 Thread Andy Townsend

On 10/11/2020 10:55, Jon Pennycook wrote:
Returning to this subject, but not necessarily at roundabouts - turn 
restrictions are still being added even where they don't exist 
(apparently) - e.g. https://www.openstreetmap.org/changeset/93759133 
 and 
https://www.openstreetmap.org/changeset/93750062 



I've commented on the first of these with a DWG hat on.

The contributor here adding turn restrictions that allegedly don't exist 
appears to be from https://www.sabre-roads.org.uk/ . They've been active 
across several areas of England; it might be worth locals taking a look 
at some of the others to check that they match on-the-ground reality.


Best Regards,

Andy


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


[Talk-GB] Turn Restrictions at roundabouts

2020-11-10 Thread Jon Pennycook
Returning to this subject, but not necessarily at roundabouts - turn
restrictions are still being added even where they don't exist (apparently)
- e.g. https://www.openstreetmap.org/changeset/93759133 and
https://www.openstreetmap.org/changeset/93750062
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-04 Thread Richard Fairhurst
Could I suggest that, rather than second-guessing what some putative router 
might or might not do, people actually try these scenarios with one of the many 
real-world routers to see if they actually happen?

I see an awful lot of "may" and "might" in this thread, together with a liberal 
sprinkling of unsourced assertions, but no actual evidence that any router has 
ever sent anyone round one of these roundabout flares.

Richard
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-04 Thread Martin Wynne

Surely OpenStreetMAP is about creating a MAP? The clue is in the name.

So you map what's on the ground:

1. Put the flares in the right place (often they are wildly out).

2. If there is a "No U-turn" sign you add it, otherwise you don't.

Making a router work properly is a job for the person making the router, 
not the person making the MAP.


Martin.

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-04 Thread James Derrick

On 04/10/2020 07:42, Edward Bainton wrote:
I've been marking them as false positives as to my mind it's obvious 
that you wouldn't U-turn there (but equally, it would be legal to do so).


You would also probably need a vehicle with a tight turning circle, and 
a quiet road - so agree it would be an unusual choice. Agree false 
positives.


But the points about machine routing make me think maybe I shouldn't 
be closing these off? Any thoughts?


Your example looks like a very common roundabout flare 3-way node with a 
cycleway crossing node. Taking the JOSM code as an example, perhaps the 
"ImproveOSM" validation also has a segment length test (<20m?) so the 
crossing is suggesting a more complex junction?



Eg, at node https://www.openstreetmap.org/node/26187838


Personally, adding restrictions to this common a use-case should be 
unnecessary (fix once in the router, not every roundabout on the map). 
Save your effort for more complex junctions where software 
understandably needs a helping hand with real-world oddities.


Happy Mapping,


James
--
James Derrick
li...@jamesderrick.org, Cramlington, England
I wouldn't be a volunteer if you paid me...
https://www.openstreetmap.org/user/James%20Derrick


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-04 Thread Edward Bainton
I regularly get QA messages about this when I use the "ImproveOSM" in iD -
just about every roundabout near me has at least one.

I've been marking them as false positives as to my mind it's obvious that
you wouldn't U-turn there (but equally, it would be legal to do so).

But the points about machine routing make me think maybe I shouldn't be
closing these off? Any thoughts?

Eg, at node https://www.openstreetmap.org/node/26187838

Missing Turn Restriction
Description
0 of 326 recorded trips (travelling west) make a turn from Bretton
Gate to Bretton
Gate at this node. There may be a missing "no_right_turn" restriction.




On Sat, 3 Oct 2020 at 20:01, James Derrick  wrote:

> Hi Dave,
>
> Thanks for answering my original router logic question! :-)
>
>
> On 03/10/2020 17:52, Dave F via Talk-GB wrote:
> > I've just tested in JOSM. It flagged no such validation warning.
> > https://www.openstreetmap.org/way/3403352
>
> Interesting - you're right, I couldn't easily reproduce the 'Sharp
> Angle' validation warning in the latest JOSM either.
>
> After hunting out the code, the warning currently isn't triggered unless
> the segment leading to a <45deg angle is <10m:
>
> https://josm.openstreetmap.de/browser/josm/trunk/src/org/openstreetmap/josm/data/validation/tests/SharpAngles.java
>
> Looking at a couple of local roundabouts via imagery, a flare this short
> verges on a single node highway=mini_roundabout, unless lots of extra
> nodes have been added to the flare to give a curved approach.
>
> After over a dozen years of using JOSM, it still surprises me with extra
> features.
>
> Happy Mapping,
>
>
> James
> --
> James Derrick
>  li...@jamesderrick.org, Cramlington, England
>  I wouldn't be a volunteer if you paid me...
>  https://www.openstreetmap.org/user/James%20Derrick
>
>
> ___
> Talk-GB mailing list
> Talk-GB@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-gb
>
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread James Derrick

Hi Dave,

Thanks for answering my original router logic question! :-)


On 03/10/2020 17:52, Dave F via Talk-GB wrote:

I've just tested in JOSM. It flagged no such validation warning.
https://www.openstreetmap.org/way/3403352


Interesting - you're right, I couldn't easily reproduce the 'Sharp 
Angle' validation warning in the latest JOSM either.


After hunting out the code, the warning currently isn't triggered unless 
the segment leading to a <45deg angle is <10m:

https://josm.openstreetmap.de/browser/josm/trunk/src/org/openstreetmap/josm/data/validation/tests/SharpAngles.java

Looking at a couple of local roundabouts via imagery, a flare this short 
verges on a single node highway=mini_roundabout, unless lots of extra 
nodes have been added to the flare to give a curved approach.


After over a dozen years of using JOSM, it still surprises me with extra 
features.


Happy Mapping,


James
--
James Derrick
li...@jamesderrick.org, Cramlington, England
I wouldn't be a volunteer if you paid me...
https://www.openstreetmap.org/user/James%20Derrick


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Dave F via Talk-GB


About the only change I've made in years of mapping 'non-mini' 
roundabouts is to split the oneway=yes flare 'V' into two segments. 
JOSM validation started flagging the junction node of the V as too 
tight a bend, which I suppose makes sense.


This is a good example of how routing is misunderstood.

A continuous way does not imply it should be the preferred route. 
Likewise, splitting a way doesn't inhibit routing that direction either; 
it would still be a "sharp angle".


I've just tested in JOSM. It flagged no such validation warning.
https://www.openstreetmap.org/way/3403352

DaveF


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Philip Barnes
On Sat, 2020-10-03 at 18:36 +0200, Colin Smale wrote:
> On 2020-10-03 18:16, Tom Hughes via Talk-GB wrote:
> > On 03/10/2020 16:57, Philip Barnes wrote:
> > 
> > 
> > > They are intended to stop this type of routing
> > > https://www.openstreetmap.org/directions?engine=graphhopper_car=52.64994%2C-1.20491%3B52.64983%2C-1.2049
> > > 
> > >  Which is techincally not illegal and in real world usage is not
> > > going to happen.
> > 
> > But unless the start or end point is on the flare why would a
> > router do that over the shorter route on the roundabout... I mean
> > maybe there are a few cases where the flare is shorter somehow?
> >  
> 
> If you take that exit by accident, a fast router may tell you to take
> the sharp turn back to the roundabout. Some flares are longer than
> others, and some routers take longer than others to trigger the "off-
> route" stuff.
>  
> 
But then is it actually illegal?
I am begining to wonder if these are mostly attempts to make QA tool
tight turn warnings go away?
Phil (trigpoint)
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Colin Smale
On 2020-10-03 18:16, Tom Hughes via Talk-GB wrote:

> On 03/10/2020 16:57, Philip Barnes wrote:
> 
>> They are intended to stop this type of routing
>> https://www.openstreetmap.org/directions?engine=graphhopper_car=52.64994%2C-1.20491%3B52.64983%2C-1.2049
>> 
>> Which is techincally not illegal and in real world usage is not going to 
>> happen.
> 
> But unless the start or end point is on the flare why would a
> router do that over the shorter route on the roundabout... I mean
> maybe there are a few cases where the flare is shorter somehow?

If you take that exit by accident, a fast router may tell you to take
the sharp turn back to the roundabout. Some flares are longer than
others, and some routers take longer than others to trigger the
"off-route" stuff.___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Tom Hughes via Talk-GB

On 03/10/2020 16:57, Philip Barnes wrote:


They are intended to stop this type of routing
https://www.openstreetmap.org/directions?engine=graphhopper_car=52.64994%2C-1.20491%3B52.64983%2C-1.2049

Which is techincally not illegal and in real world usage is not going to 
happen.


But unless the start or end point is on the flare why would a
router do that over the shorter route on the roundabout... I mean
maybe there are a few cases where the flare is shorter somehow?

Tom

--
Tom Hughes (t...@compton.nu)
http://compton.nu/

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Philip Barnes
On Sat, 2020-10-03 at 16:50 +0100, Tom Hughes via Talk-GB wrote:
> On 03/10/2020 14:05, Brian Prangle wrote:
> 
> > There seems to be a predilection for adding turn restrictions ,
> > either 
> > no right rurns or no U turns at the exit flares of roundabouts to 
> > prevent turning back into the entry flares where there are no
> > explicit 
> > signed restrictions. I suspect this is "rendering for routers". Do 
> > routers actually need this data?  I'm tempted just to delete them
> > all 
> > wherever I meet them, but I suspect there are thousands of them
> > and 
> > there'll be howls of complaint.
> 
> Surely if there is a flare then the entry half of the flare will
> be one-way against anybody turning off which should exclude it from

Tom

They are intended to stop this type of routing
https://www.openstreetmap.org/directions?engine=graphhopper_car=52.64994%2C-1.20491%3B52.64983%2C-1.2049

Which is techincally not illegal and in real world usage is not going
to happen.

Phil (trigpoint)
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread James Derrick

Hi,

On 03/10/2020 14:05, Brian Prangle wrote:
There seems to be a predilection for adding turn restrictions , either 
no right rurns or no U turns at the exit flares of roundabouts to 
prevent turning back into the entry flares where there are no explicit 
signed restrictions. I suspect this is "rendering for routers". Do 
routers actually need this data?  I'm tempted just to delete them all 
wherever I meet them, but I suspect there are thousands of them and 
there'll be howls of complaint.


About the only change I've made in years of mapping 'non-mini' 
roundabouts is to split the oneway=yes flare 'V' into two segments. JOSM 
validation started flagging the junction node of the V as too tight a 
bend, which I suppose makes sense.


I wonder if seeing a junction node of three vertices rather than a 
300degree turn on one segment make a difference to roundabout unaware 
routers?


Poor attempt at an ASCII diagram explaining what a two segment flare is 
below!


Flare 1 ->---\

              * Road 3 continues away from roundabout

Flare 2 -<---/

Flares are the oneway=yes roads connecting to the roundabout.
These are often a single V shaped segment, rather than two plus the road.


James
--
James Derrick
li...@jamesderrick.org, Cramlington, England
I wouldn't be a volunteer if you paid me...
https://www.openstreetmap.org/user/James%20Derrick


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Tom Hughes via Talk-GB

On 03/10/2020 14:05, Brian Prangle wrote:

There seems to be a predilection for adding turn restrictions , either 
no right rurns or no U turns at the exit flares of roundabouts to 
prevent turning back into the entry flares where there are no explicit 
signed restrictions. I suspect this is "rendering for routers". Do 
routers actually need this data?  I'm tempted just to delete them all 
wherever I meet them, but I suspect there are thousands of them and 
there'll be howls of complaint.


Surely if there is a flare then the entry half of the flare will
be one-way against anybody turning off which should exclude it from
any consideration by a router?

Tom

--
Tom Hughes (t...@compton.nu)
http://compton.nu/

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Brian Prangle
Where I've seen them they're on every exit

On Sat, 3 Oct 2020, 15:49 Dave F via Talk-GB, 
wrote:

> There is a lot of garbage in OSM due to those creating routers being too
> lazy to write a few lines of code, or even use common sense.
>
> However, in this case I believe it's other contributors who think, for
> reasons that escape me, routers require it.
>
> Routing software must be of a poor standard if it returns the commuter to
> the roundabout after just 10 metres or so after leaving it.
>
>  I partially agree with Phil's suggestion that they're harmless, if turn
> restrictions are by themselves. However if there are  multiple at one
> junction it can become error prone. Any that aren't required are best
> removed.
>
> DaveF
>
>
> On 03/10/2020 14:05, Brian Prangle wrote:
>
> Hi
>
> There seems to be a predilection for adding turn restrictions , either no
> right rurns or no U turns at the exit flares of roundabouts to prevent
> turning back into the entry flares where there are no explicit signed
> restrictions. I suspect this is "rendering for routers". Do routers
> actually need this data?  I'm tempted just to delete them all wherever I
> meet them, but I suspect there are thousands of them and there'll be howls
> of complaint.
>
> Regards
>
> Brian
>
> ___
> Talk-GB mailing 
> listTalk-GB@openstreetmap.orghttps://lists.openstreetmap.org/listinfo/talk-gb
>
>
> ___
> Talk-GB mailing list
> Talk-GB@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-gb
>
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Dave F via Talk-GB
There is a lot of garbage in OSM due to those creating routers being too 
lazy to write a few lines of code, or even use common sense.


However, in this case I believe it's other contributors who think, for 
reasons that escape me, routers require it.


Routing software must be of a poor standard if it returns the commuter 
to the roundabout after just 10 metres or so after leaving it.


 I partially agree with Phil's suggestion that they're harmless, if 
turn restrictions are by themselves. However if there are  multiple at 
one junction it can become error prone. Any that aren't required are 
best removed.


DaveF


On 03/10/2020 14:05, Brian Prangle wrote:

Hi

There seems to be a predilection for adding turn restrictions , either 
no right rurns or no U turns at the exit flares of roundabouts to 
prevent turning back into the entry flares where there are no explicit 
signed restrictions. I suspect this is "rendering for routers". Do 
routers actually need this data?  I'm tempted just to delete them all 
wherever I meet them, but I suspect there are thousands of them and 
there'll be howls of complaint.


Regards

Brian

___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


Re: [Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Philip Barnes
On Sat, 2020-10-03 at 14:05 +0100, Brian Prangle wrote:
> Hi
> 
> There seems to be a predilection for adding turn restrictions ,
> either no right rurns or no U turns at the exit flares of roundabouts
> to prevent turning back into the entry flares where there are no
> explicit signed restrictions. I suspect this is "rendering for
> routers". Do routers actually need this data?  I'm tempted just to
> delete them all wherever I meet them, but I suspect there are
> thousands of them and there'll be howls of complaint. 
> 
They do seem to be unnecessary and are not legal unless signed.

Whilst you may find them playing with routers on a pc, no real world
satnav will ever detect that you have taken a wrong  turning and
reroute within a timeframe that these would ever be an issue.

I would probably remove them after survey and it could be worth
commenting when mapper adds these asking for reasoning. 

The danger of removing them without survey is that we could remove a
rare real restriction. They are unecessary but harmless IMHO.

Phil (trigpoint)
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb


[Talk-GB] Turn Restrictions at roundabouts

2020-10-03 Thread Brian Prangle
Hi

There seems to be a predilection for adding turn restrictions , either no
right rurns or no U turns at the exit flares of roundabouts to prevent
turning back into the entry flares where there are no explicit signed
restrictions. I suspect this is "rendering for routers". Do routers
actually need this data?  I'm tempted just to delete them all wherever I
meet them, but I suspect there are thousands of them and there'll be howls
of complaint.

Regards

Brian
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb