[OSM-talk] Rendering of footways with bicycle=yes

Fri May 1 11:10:43 BST 2009

Ed

I'm not surprised you're confused - there is indeed a great deal of
confusion! So everyone is likely to give you a different answer.

As someone who has lived for a good number of years both in the USA and in
Europe (both England and Germany) I can relate to your problems with cycling
in the USA!

In my experience - such as it is - practice in OSM (with regard to
non-motorised highways) tends to start out with what makes very good sense
in Germany and then gets modified to take account of what makes sense in the
UK before being further modified by inputs from the USA and elsewhere. The
UK system is almost unique inasmuch as there is a very clearly laid down
legal set of definitions for "public rights of way" - but unfortunately the
legal situation does not always correspond to what is 'common sense' in
terms of customary usage or what you see 'on the ground'.

I tend to think that the tag highway=cycleway is used in three broadly
different situations: (a) where there is a dedicated lane alongside a
vehicular highway - separated from it or not - that is designed either
exclusively (usually) for cyclists or occasionally jointly with pedestrians;
(b) where there is a dedicated way - usually paved and in an urban area -
that is designated by signage as being for cyclists only or for cyclists and
pedestrians - the way may also be split by a painted road marking into two
parallel lanes (separated only by the marking) - one each for cyclists
(only) and the other for pedestrians (only); (c) where there is a way -
usually paved and usually in rural areas - that has been created primarily
with cyclists in mind (but which - in the UK at least to the best of my
knowledge - is also available for use by pedestrians.

Given that situation the potential for confusion is obviously high and the
tagging tends a bit to depend on whether it was done by someone who is
mostly a cyclist or someone who is mostly a walker.

If you scan this list you will find a lot of discussion as to the pros and
cons of various approaches - the wiki also gives somewhat varying clues
according to the latest edit!

At present - although I have modified my own approach several times in the
light of comments from others on this list - I tend to use highway=cycleway
ONLY for the three cases listed above. Other people might use it more
widely. I tend to modify tagging with things like
foot/bicycle=no/permissive/yes where some clarification would seem helpful.
In the UK I also use designation= to reflect legal status without regard to
the physical condition on the ground (which can either be reflected in the
main highway= tag or by adding tags such as surface= or tracktype= ).

I don't think any of this is going to be resolved completely in the near
future so I am sure most people would say 'feel free to use your own best
judgement'!

Mike Harris

  _____  

From: Hillsman, Edward [mailto:hillsman at cutr.usf.edu] 
Sent: 30 April 2009 19:40
To: talk at openstreetmap.org
Subject: Re: [OSM-talk] Rendering of footways with bicycle=yes

I'm one of the people mapping paths (since March) who scans this list, and I
have to say that I'm confused. Although part of that may be because I'm new
to OSM and not just to the matter of how to deal with tagging and rendering
things. And part of that may because a lot of the tagging conventions
developed in Europe, where the cycling infrastructure is often much better
than in most of the United States.

I got into OSM because I think it and its associated community of spin-off
applications provide the best opportunity for most communities in the US to
enable citizens to generate routes so that they can plan trips by bicycle.
The cycling infrastructure in most parts of the US is discontinuous, poorly
mapped by public agencies, and consists of a mix of types:  shoulders along
roads designated as bike lanes (no curb to the outside); similar but
undesignated shoulders that cyclists discover but are not "official"; lanes
marked within streets, often adjacent to outside curbs, but sometimes
between lanes of motor-vehicle traffic; sidewalks (footways) parallel to
major streets, which were built with the intent of being used by cyclists;
traditional sidewalks that were not but which may be used by cyclists except
where prohibited; dedicated paths/trails built separate from the road
right-of-way, which may be used for utilitarian travel but which often are
located where they are used primarily for recreation rather than "real"
trips (most of which are designated "multi-use" and are used by cyclists and
pedestrians); and the majority of roads, which cyclists are legally entitled
to use, but which are not specially marked, and which may or may not be
unsafe to ride.

It is common to have cycling infrastructure on one side of a street but not
the other; some types may be safe for two-way cycling, but others, such as
shoulders and most in-street lanes, definitely are not. Where the street is
divided by a median, as in a boulevard, it is easy to code the street as two
one-way paths, code the cycling infrastructure separately on each, and let
the oneway=yes tag take care of this. Where the street is a two-lane,
two-way street with a shoulder or lane on one side, clearly intended to be
used in one direction and not the other and no cycling infrastructure on the
other side of the street, there is a problem. This is common in Tampa, and I
welcome guidance.

Some questions about coding:

I assume that highway=cycleway is a path developed outside a road
right-of-way, primarily for cycling (and the topic that you have been
discussing in this thread). The illustration on the Map Features page lacks
enough surrounding context to indicate whether the tag might be suitable for
other kinds of cycling infrastructure. If I am correct, then what would be
the difference between this and cycleway=track?

Cycleway=lane, the illustration shows what could either be a bicycle
shoulder or an in-street bicycle lane. These have very different perceptual
"feel" to cyclists, depending on the character of the main road, the motor
traffic on it, the volume and speed of the motor traffic, and the geometry
of the lane or shoulder. On one street here, there is a lane (officially,
"excellent" cycling infrastructure) which most cyclists veer out of to use
the shoulder instead, which at that point is not designated as a bike
shoulder, because there is a lane. If you saw the section of street, you
would understand why.

Cycleway=track would cover the multi-use, largely recreational,
infrastructure. It might or might not be intended for the sidewalks intended
to be used by cyclists.

Cycleway=opposite_lane is rare here, and in the US is probably only suitable
for low-volume streets except in areas with large numbers of cyclists, such
as Portland, Davis, or Boulder. See below.

Cycleway=opposite_track again might or might not be intended for the
sidewalks intended to be used by cyclists, which often are on just one side
of the street. Unfortunately, research has demonstrated these to be
dangerous when cyclists who use them against the flow of motor traffic must
cross an intersection (because drivers are not looking for them there).

I have attempted to tag some of the multi-use paths as highway=footway and
as highway=cycleway, but only the most recently entered survives. Most of
the multi-use paths with which I am familiar have been entered by others and
tagged as highway=footway. What is the best way to designate their
multipurpose character? I assume add bicycle=yes.

Thinking ahead toward the objective of having routing algorithm available to
use this to generate bicycle routes, how can we code these various types in
ways that someone can eventually make usable routes out of them? 

If you are aware of anyone developing such a routing facility to run using
data from OSM, could you refer me to him/her?

At the moment, many large cities in the US have no OSM mapping activity at
all, and in most of those which do, it is fragmentary. It would be very good
to get this sorted out before lots more people here become involved.

Ed Hillsman

>On Thu, 30 Apr 2009 16:59:50 +0100, Andy Allen

>gravitystorm at gmail.com wrote:

> 

>Subject: Re: [OSM-talk] Rendering of footways with bicycle=yes

>To: Richard Mann <richard.mann.westoxford at googlemail.com>

>Cc: talk at openstreetmap.org

>Message-ID:

>     <c4193f8c0904300859w5129fc28pdbc264c08c92039f at mail.gmail.com>

>Content-Type: text/plain; charset=ISO-8859-1

> 

>On Thu, Apr 30, 2009 at 3:10 PM, Richard Mann

><richard.mann.westoxford at googlemail.com> wrote:

>> I'd support that highway=path needs to be rendered in the cycle map
layer,

>> especially now it's becoming clearer how it's being used

> 

>Every time it gets discussed, it becomes *less* clear how it's being

>used to me. And I'm mightily concerned that the 10 people discussing

>it on these lists might be in no way representative of the 14,990

>people who are mapping paths and aren't in these discussions.

> 

>Cheers,

>Andy

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20090501/c198b3de/attachment.html>