[talk-au] Local bicycle routes in NSW

[Tom Brennan]

Willoughby Council (Sydney, NSW) has recently been refreshing its cycle 
route signage, so I've been riding the routes and reviewing tagging in 
OSM. Before I go and make a whole lot of changes, I just wanted to 
confirm best practice.


1. Infrastructure:
Painted road markings (but no cycle lane) and/or street signs indicating 
cycle route:

cycleway=shared_lane
eg 
https://ozultimate.com/temp/2020-04-24%2022_01_11-NSW%20Bushwalking%20Maps.png


Painted cycle lane:
cycleway=lane
https://ozultimate.com/temp/2020-04-24%2022_03_56-NSW%20Bushwalking%20Maps.png

This seems to be my interpretation of:
https://wiki.openstreetmap.org/wiki/Key:cycleway
and
https://wiki.openstreetmap.org/wiki/Australian_Tagging_Guidelines


2. Cycle Routes:
Use of network=lcn vs lcn=yes - I assume network=lcn is preferred to 
lcn=yes? Quite a lot of the current routes have lcn=yes.


Alternatively, should I be trying to create relations? The problem with 
relations is that the cycleways all interconnect. So while there may be 
a sensible route from any suburb in Willoughby to any other suburb, but 
it doesn't seem to lend itself to a collection of relations. Certainly 
the signage at any given spot just points you to the next suburb.


The link below shows the approximate network (not all are yet 
built/marked - I'll be updating OSM on the basis of ground surveys)

http://edocs.willoughby.nsw.gov.au/DocumentViewer.ashx?dsi=2914874
I don't see that it easily lends itself to relations.


cheers
Tom

Canyoning? try http://ozultimate.com/canyoning
Bushwalking? try http://bushwalkingnsw.com

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

Re: [talk-au] Local bicycle routes in NSW

[Warin]

On 25/4/20 8:55 am, Tom Brennan wrote:
Willoughby Council (Sydney, NSW) has recently been refreshing its 
cycle route signage, so I've been riding the routes and reviewing 
tagging in OSM. Before I go and make a whole lot of changes, I just 
wanted to confirm best practice.


1. Infrastructure:
Painted road markings (but no cycle lane) and/or street signs 
indicating cycle route:

cycleway=shared_lane
eg 
https://ozultimate.com/temp/2020-04-24%2022_01_11-NSW%20Bushwalking%20Maps.png


Painted cycle lane:
cycleway=lane
https://ozultimate.com/temp/2020-04-24%2022_03_56-NSW%20Bushwalking%20Maps.png 



This seems to be my interpretation of:
https://wiki.openstreetmap.org/wiki/Key:cycleway
and
https://wiki.openstreetmap.org/wiki/Australian_Tagging_Guidelines


2. Cycle Routes:
Use of network=lcn vs lcn=yes - I assume network=lcn is preferred to 
lcn=yes? Quite a lot of the current routes have lcn=yes.


Alternatively, should I be trying to create relations? The problem 
with relations is that the cycleways all interconnect. So while there 
may be a sensible route from any suburb in Willoughby to any other 
suburb, but it doesn't seem to lend itself to a collection of 
relations. Certainly the signage at any given spot just points you to 
the next suburb.


The link below shows the approximate network (not all are yet 
built/marked - I'll be updating OSM on the basis of ground surveys)

http://edocs.willoughby.nsw.gov.au/DocumentViewer.ashx?dsi=2914874
I don't see that it easily lends itself to relations.



My opinion.

Routes go from A to B. They are not simple road segments.

An example?

Relation: Northbridge-Castle Cove (6282327)
  Tags:
    "name"="Northbridge-Castle Cove"
    "ref"="NCC"
    "route"="bicycle"
    "type"="route"
    "lcn"="yes"
    "network"="Willoughby"

The above is correct.

It contains numerous road segments (ways). Some of these are tagged 
lcn=yes. This is wrong.


Example?

Way: Baringa Road (794266238)
  Tags:
    "source:name"="historical"
    "surface"="paved"
    "maxspeed"="50"
    "name"="Baringa Road"
    "source"="yahoo_imagery"
    "highway"="residential"
    "cycleway"="shared_lane"
    "network"="lcn"

There should be no   "network"="lcn" on the as it does not, by itself, 
form a route.


Similarly I would remove the tag "lcn=yes" on any simple way.

(I would also remove the source tag - I would assume that the state 
source is old and there would have been a few edits of this from other 
sources in the mean time.)



Some routes will use the same ways as another route. That is fine, happens.

Some bicycle routes use parts of hiking routes, sometimes bus routes etc 
etc. This too is fine.





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

Re: [talk-au] Local bicycle routes in NSW

[Andrew Harvey]

On Sat, 25 Apr 2020 at 08:57, Tom Brennan  wrote:

> Willoughby Council (Sydney, NSW) has recently been refreshing its cycle
> route signage, so I've been riding the routes and reviewing tagging in
> OSM. Before I go and make a whole lot of changes, I just wanted to
> confirm best practice.


> 1. Infrastructure:
> Painted road markings (but no cycle lane) and/or street signs indicating
> cycle route:
> cycleway=shared_lane
> eg
>
> https://ozultimate.com/temp/2020-04-24%2022_01_11-NSW%20Bushwalking%20Maps.png
>

cycleway=shared_lane is only if there is a marking on the ground, so if
it's a marked route but no bicycle stencil on the ground then it's not
cycleway=shared_lane in my view because cycleway is tagging the
infrastructure on the ground.

Painted cycle lane:
> cycleway=lane
>
> https://ozultimate.com/temp/2020-04-24%2022_03_56-NSW%20Bushwalking%20Maps.png


Correct.

>
>
> This seems to be my interpretation of:
> https://wiki.openstreetmap.org/wiki/Key:cycleway
> and
> https://wiki.openstreetmap.org/wiki/Australian_Tagging_Guidelines
>
>
> 2. Cycle Routes:
> Use of network=lcn vs lcn=yes - I assume network=lcn is preferred to
> lcn=yes? Quite a lot of the current routes have lcn=yes.
>

network=lcn goes on the relation not on the way. lcn=yes goes on the way,
but is redundant if it's already part of a relation. Personally I'd use
lcn=yes on a way if I know the segment is part of a route, but don't know
or have time to map out the full route relation.


>
> Alternatively, should I be trying to create relations? The problem with
> relations is that the cycleways all interconnect. So while there may be
> a sensible route from any suburb in Willoughby to any other suburb, but
> it doesn't seem to lend itself to a collection of relations. Certainly
> the signage at any given spot just points you to the next suburb.
>

Yeah, that's why I don't usually create full route relations unless it's
clear what the route is, if there are signs directing cyclists in various
directions then it's either a lcn/rcn/ncn=yes on the way or put it in a
route relation.


> The link below shows the approximate network (not all are yet
> built/marked - I'll be updating OSM on the basis of ground surveys)
> http://edocs.willoughby.nsw.gov.au/DocumentViewer.ashx?dsi=2914874
> I don't see that it easily lends itself to relations.
___
Talk-au mailing list
Talk-au@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au

Re: [talk-au] Local bicycle routes in NSW

[Andrew Harvey]

On Sat, 25 Apr 2020 at 10:31, Warin <61sundow...@gmail.com> wrote:

> My opinion.
>
> Routes go from A to B. They are not simple road segments.
>
> An example?
>
> Relation: Northbridge-Castle Cove (6282327)
>Tags:
>  "name"="Northbridge-Castle Cove"
>  "ref"="NCC"
>  "route"="bicycle"
>  "type"="route"
>  "lcn"="yes"
>  "network"="Willoughby"
>
> The above is correct.
>
> It contains numerous road segments (ways). Some of these are tagged
> lcn=yes. This is wrong.
>

Agreed, the lcn=yes should go on the way segment (but is redundant if a
relation exists) and not on the relation.


>
> Example?
>
> Way: Baringa Road (794266238)
>Tags:
>  "source:name"="historical"
>  "surface"="paved"
>  "maxspeed"="50"
>  "name"="Baringa Road"
>  "source"="yahoo_imagery"
>  "highway"="residential"
>  "cycleway"="shared_lane"
>  "network"="lcn"
>
> There should be no   "network"="lcn" on the as it does not, by itself,
> form a route.
>

Agreed, the network tag should go on the relation not the way, the way
itself could have lcn=yes if you know it's part of a route but it hasn't
been mapped out as a relation yet.


> Similarly I would remove the tag "lcn=yes" on any simple way.
>

Only if it's part of a network=lcn relation already, if not it's still
useful to say there is a route here, but the route hasn't yet been mapped
out as a relation.


>
> (I would also remove the source tag - I would assume that the state
> source is old and there would have been a few edits of this from other
> sources in the mean time.)
>

I only do that if I've changed the source, so if it's refering to geometry
and I've updated the geometry I'll update it, but if you've made
significant changes then I also think it's fair to remove since it's still
present in the history.
___
Talk-au mailing list
Talk-au@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au

Re: [talk-au] Local bicycle routes in NSW

[Warin]

On 25/4/20 11:50 am, Andrew Harvey wrote:



On Sat, 25 Apr 2020 at 10:31, Warin <61sundow...@gmail.com 
> wrote:


My opinion.

Routes go from A to B. They are not simple road segments.

An example?

Relation: Northbridge-Castle Cove (6282327)
   Tags:
 "name"="Northbridge-Castle Cove"
 "ref"="NCC"
 "route"="bicycle"
 "type"="route"
 "lcn"="yes"
 "network"="Willoughby"

The above is correct.



Looking at it again ...

"network"="lcn" appears to be more correct...

Possibly "Willoughby" could be tagged "cycle_network"="Willoughby"   see 
https://wiki.openstreetmap.org/wiki/Cycle_routes#Relations






It contains numerous road segments (ways). Some of these are tagged
lcn=yes. This is wrong.


Agreed, the lcn=yes should go on the way segment (but is redundant if 
a relation exists) and not on the relation.



Example?

Way: Baringa Road (794266238)
   Tags:
 "source:name"="historical"
 "surface"="paved"
 "maxspeed"="50"
 "name"="Baringa Road"
 "source"="yahoo_imagery"
 "highway"="residential"
 "cycleway"="shared_lane"
 "network"="lcn"

There should be no   "network"="lcn" on the as it does not, by
itself,
form a route.


Agreed, the network tag should go on the relation not the way, the way 
itself could have lcn=yes if you know it's part of a route but it 
hasn't been mapped out as a relation yet.


Similarly I would remove the tag "lcn=yes" on any simple way.


Only if it's part of a network=lcn relation already, if not it's still 
useful to say there is a route here, but the route hasn't yet been 
mapped out as a relation.



I would start the relation rather than use "lcn=yes". Then, in the 
relation, a source can be stated (together with any comments, 
description etc) that is clearly the source of the route at this stage.


This would also cause it to show up on 
https://cycling.waymarkedtrails.org/#route?id=172 in a meaning full 
way.


Note that this route is incomplete.

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

Re: [talk-au] Local bicycle routes in NSW

[Tom Brennan]

On 25/04/2020 11:46 am, Andrew Harvey wrote:

On Sat, 25 Apr 2020 at 08:57, Tom Brennan 
wrote:

1. Infrastructure: Painted road markings (but no cycle lane) and/or
street signs indicating cycle route: cycleway=shared_lane eg

https://ozultimate.com/temp/2020-04-24%2022_01_11-NSW%20Bushwalking%20Maps.png


cycleway=shared_lane is only if there is a marking on the ground, so
if it's a marked route but no bicycle stencil on the ground then it's
not cycleway=shared_lane in my view because cycleway is tagging the 
infrastructure on the ground.


Agree in general.

The one exception is that many of the cycle routes dogleg through the 
residential streets. In some cases, I don't think the short legs of the 
doglegs have bicycle stencils on the ground as they are too short. But 
they do have cycle street signage (directional arrows), so I'd argue 
that these are intended to be shared lanes.


Example:
https://ozultimate.com/temp/2020_04_25_12_38_06_NSW_Bushwalking_Maps.png
The red legs have stencils on the ground, the blue legs (from memory) 
have street signage.


>> 2. Cycle Routes: Use of network=lcn vs lcn=yes - I assume
>> network=lcn is preferred to lcn=yes? Quite a lot of the current
>> routes have lcn=yes.
>
> network=lcn goes on the relation not on the way. lcn=yes goes on the
> way, but is redundant if it's already part of a relation. Personally
> I'd use lcn=yes on a way if I know the segment is part of a route,
> but don't know or have time to map out the full route relation.

Ah, I think that generally makes it clearer.

On the relation, what about network=Willoughby vs network=lcn? CycleOSM 
seems to be expecting network=lcn, not that tagging for a renderer is key!


A couple of route examples I can see:
https://www.openstreetmap.org/relation/6282327
https://www.openstreetmap.org/relation/6282327
have network=Willoughby and lcn=yes.

However, if I go over to Cammeray, someone has added all of the ways to 
a single relation (named Cammeray Local Routes, tagged with lcn=yes and 
network=lcn). In some ways this makes sense, since these reflect the 
interconnected series of local routes. It's not really a single route, 
but it is a useful way of grouping.

https://www.openstreetmap.org/relation/6283487
Naremburn is similar:
https://www.openstreetmap.org/relation/6282427


The references I assume are just made up to give renderers something to 
put on the various maps. They don't seem to relate to anything on any 
signs or in council brochures.


cheers
Tom

Canyoning? try http://ozultimate.com/canyoning
Bushwalking? try http://bushwalkingnsw.com

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

Re: [talk-au] Local bicycle routes in NSW

[Andrew Harvey]

On Sat, 25 Apr 2020 at 15:13, Tom Brennan  wrote:

> On 25/04/2020 11:46 am, Andrew Harvey wrote:
> > On Sat, 25 Apr 2020 at 08:57, Tom Brennan 
> > wrote:
> >> 1. Infrastructure: Painted road markings (but no cycle lane) and/or
> >> street signs indicating cycle route: cycleway=shared_lane eg
> >>
> >>
> https://ozultimate.com/temp/2020-04-24%2022_01_11-NSW%20Bushwalking%20Maps.png
> >
> > cycleway=shared_lane is only if there is a marking on the ground, so
> > if it's a marked route but no bicycle stencil on the ground then it's
> > not cycleway=shared_lane in my view because cycleway is tagging the
> > infrastructure on the ground.
>
> Agree in general.
>
> The one exception is that many of the cycle routes dogleg through the
> residential streets. In some cases, I don't think the short legs of the
> doglegs have bicycle stencils on the ground as they are too short. But
> they do have cycle street signage (directional arrows), so I'd argue
> that these are intended to be shared lanes.
>
> Example:
> https://ozultimate.com/temp/2020_04_25_12_38_06_NSW_Bushwalking_Maps.png
> The red legs have stencils on the ground, the blue legs (from memory)
> have street signage.
>

The way cycleway=shared_lane was proposed and documented (
https://wiki.openstreetmap.org/wiki/Proposed_features/shared_lane) as a
painted icon on the road which indicates to both drivers and cyclists that
the road is shared, even if in Australia it's always shared regardless of
signage, I do think it only makes sense to tag it to indicate the painted
signage on the road. I don't think a beware of cylists or look out for
cyclists road sign counts as cycleway=shared_lane.

I can't see any markings on my Mapillary imagery
https://www.mapillary.com/map/im/lpdzCigIMdASq4IzX248lA but that's only
from 2017.


>
>  >> 2. Cycle Routes: Use of network=lcn vs lcn=yes - I assume
>  >> network=lcn is preferred to lcn=yes? Quite a lot of the current
>  >> routes have lcn=yes.
>  >
>  > network=lcn goes on the relation not on the way. lcn=yes goes on the
>  > way, but is redundant if it's already part of a relation. Personally
>  > I'd use lcn=yes on a way if I know the segment is part of a route,
>  > but don't know or have time to map out the full route relation.
>
> Ah, I think that generally makes it clearer.
>
> On the relation, what about network=Willoughby vs network=lcn? CycleOSM
> seems to be expecting network=lcn, not that tagging for a renderer is key!
>

Yes see https://wiki.openstreetmap.org/wiki/Cycle_routes#Relations network
should be lcn for local routes, rcn for regional routes and ncn for
national routes.


>
> A couple of route examples I can see:
> https://www.openstreetmap.org/relation/6282327
> https://www.openstreetmap.org/relation/6282327
> have network=Willoughby and lcn=yes.
>
> However, if I go over to Cammeray, someone has added all of the ways to
> a single relation (named Cammeray Local Routes, tagged with lcn=yes and
> network=lcn). In some ways this makes sense, since these reflect the
> interconnected series of local routes. It's not really a single route,
> but it is a useful way of grouping.
> https://www.openstreetmap.org/relation/6283487
> Naremburn is similar:
> https://www.openstreetmap.org/relation/6282427


That's kind of why I don't usually bother with the route relations and
instead just tag ways as lcn=yes because there is usually no clear A to B
route and instead just a lot of connections. So in that way tagging lcn=yes
can still help routers prefer cycle routes but we don't need to bother
working out what the route is called and where it starts and ends, which
unless it's surveyable on the ground, I wouldn't bother.

In both those examples "Naremburn Local Routes" is just someones
description and not a signposted or at least well known route name, so I'd
be happy to just remove it but interested to hear what others think. I know
Warrin said he things routes should go from A to B.


>
>
>
> The references I assume are just made up to give renderers something to
> put on the various maps. They don't seem to relate to anything on any
> signs or in council brochures.
>
> cheers
> Tom
> 
> Canyoning? try http://ozultimate.com/canyoning
> Bushwalking? try http://bushwalkingnsw.com
>
> ___
> Talk-au mailing list
> Talk-au@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-au
>
___
Talk-au mailing list
Talk-au@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au

Re: [talk-au] Local bicycle routes in NSW

[Richard Fairhurst]

Tom Brennan wrote:
> However, if I go over to Cammeray, someone has added all of the ways 
> to a single relation (named Cammeray Local Routes, tagged with 
> lcn=yes and network=lcn).

Yeah, please don't do that. :)

Relations with type=route are for routes, with a defined start and end. Not
for networks. If you want to put them all in a single relation, then do it
with type=network or something:

https://wiki.openstreetmap.org/wiki/Relation:network

cheers
Richard
cycle.travel



--
Sent from: http://gis.19327.n8.nabble.com/Australia-f5416966.html

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

Re: [talk-au] Local bicycle routes in NSW

[Andrew Harvey]

On Sat, 25 Apr 2020 at 18:49, Richard Fairhurst 
wrote:

> Tom Brennan wrote:
> > However, if I go over to Cammeray, someone has added all of the ways
> > to a single relation (named Cammeray Local Routes, tagged with
> > lcn=yes and network=lcn).
>
> Yeah, please don't do that. :)
>
> Relations with type=route are for routes, with a defined start and end. Not
> for networks. If you want to put them all in a single relation, then do it
> with type=network or something:
>
> https://wiki.openstreetmap.org/wiki/Relation:network


Do you have an example of that for cycle networks?
___
Talk-au mailing list
Talk-au@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au

Re: [talk-au] Local bicycle routes in NSW

[Richard Fairhurst]

On 25 Apr 2020, 09:53 +0100, Andrew Harvey , wrote:
>
> > On Sat, 25 Apr 2020 at 18:49, Richard Fairhurst  
> > wrote:
> > > Relations with type=route are for routes, with a defined start and end. 
> > > Not
> > > for networks. If you want to put them all in a single relation, then do it
> > > with type=network or something:
> > >
> > > https://wiki.openstreetmap.org/wiki/Relation:network
> >
> > Do you have an example of that for cycle networks?

I've seen it in the US a few times - can't remember where offhand. Ultimately 
there isn't really a whole lot of point - if you want all the routes in 
Cammeray, just get a bounding box for Cammeray and find the cycle routes within 
it. OSM is a spatial database after all. But people do like categorising things!

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

Re: [talk-au] Local bicycle routes in NSW

[Andrew Harvey]

For these "routes" though there is no clear A to B, there will be short
segments which are obivously part of a route because there are arrows
directing cyclists, but sometimes these are just short segments to the next
intersection so it's unclear where the route goes from and to, hence why
someone has resorted to just dumping all the segments into one route
relation.

On Sat, 25 Apr 2020 at 18:56, Richard Fairhurst 
wrote:

> On 25 Apr 2020, 09:53 +0100, Andrew Harvey ,
> wrote:
>
>
> On Sat, 25 Apr 2020 at 18:49, Richard Fairhurst 
> wrote:
>
>> Relations with type=route are for routes, with a defined start and end.
>> Not
>> for networks. If you want to put them all in a single relation, then do it
>> with type=network or something:
>>
>> https://wiki.openstreetmap.org/wiki/Relation:network
>
>
> Do you have an example of that for cycle networks?
>
>
> I've seen it in the US a few times - can't remember where offhand.
> Ultimately there isn't really a whole lot of point - if you want all the
> routes in Cammeray, just get a bounding box for Cammeray and find the cycle
> routes within it. OSM is a spatial database after all. But people do like
> categorising things!
>
> Richard
>
___
Talk-au mailing list
Talk-au@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au

Re: [talk-au] Local bicycle routes in NSW

[Richard Fairhurst]

Andrew Harvey , wrote:
> For these "routes" though there is no clear A to B, there will be short 
> segments which are obivously part of a route because there are arrows 
> directing cyclists, but sometimes these are just short segments to the next 
> intersection so it's unclear where the route goes from and to, hence why 
> someone has resorted to just dumping all the segments into one route relation.

Exactly, so it’s not an A-B “route”, it’s a network, and should be in a network 
relation rather than a route relation.

The other alternative is to just put lcn=yes on the way (and indeed that’s done 
in lots of other places). cycle.travel gives a small uplift to ways tagged with 
that.

Richard
cycle.travel
___
Talk-au mailing list
Talk-au@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au

Re: [talk-au] Local bicycle routes in NSW

[Andrew Harvey]

On Sat, 25 Apr 2020 at 19:09, Richard Fairhurst 
wrote:

> Andrew Harvey , wrote:
>
> For these "routes" though there is no clear A to B, there will be short
> segments which are obivously part of a route because there are arrows
> directing cyclists, but sometimes these are just short segments to the next
> intersection so it's unclear where the route goes from and to, hence why
> someone has resorted to just dumping all the segments into one route
> relation.
>
>
> Exactly, so it’s not an A-B “route”, it’s a network, and should be in a
> network relation rather than a route relation.
>

But which network relation should they be part of since they can link in
all sorts of ways it's hard to say which gets grouped together.


> The other alternative is to just put lcn=yes on the way (and indeed that’s
> done in lots of other places). cycle.travel gives a small uplift to ways
> tagged with that.
>

Yep, that's my personal preference where there is no obvious named route.
___
Talk-au mailing list
Talk-au@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au