Discussion:
[OSM-talk] Path vs footway vs cycleway vs...
Steve Bennett
2009-11-28 08:24:02 UTC
Permalink
Hi all,
(Apologies if this is the wrong list - still getting my head around
them all. Or this has been discussed extensively, please point me at
it)...

I'm doing a lot of mapping of pedestrian and bike paths around my
area, and am having trouble deciding when to use path, when footway,
and when cycleway. I'm particularly troubled by the way Potlatch
describes "path" as "unofficial path" - making it sound like an
unpaved line of footprints carved through the grass.

Could someone give me guidance on a few specific scenarios:
1) In the parks near me, there are lots of paths, which I guess were
probably intended for pedestrians, but cyclists use them too.
Sometimes paved, sometimes not. I've been tagging them "highway=path,
bicycle=yes" (to be safe).

2) Multi-use paths, like in new housing developments. Usually paved,
and connecting streets together.

3) Genuine multi-use paths along the sides of creeks or freeways.
Frequently with a dotted line down the middle. Most people think of
them as bike paths, but plenty of pedestrians use them too.
"highway=cycleway, foot=yes" seems the most satisfying, but according
to the definition, it should just be a "path"? I tend to assume it's a
cycleway if the gap between two entrances ever exceeds a kilometre or
so...

4) In Albert Park (home of the grand prix) near me, there are lots of
sealed paths that are wide enough for a car. They're normally blocked
off, and used mainly by contractors before and after the grand prix.
The rest of the time, they're used by pedestrians and cyclists. I had
marked them "highway=unclassified" but now I think "highway=track
surface=paved" would be better?

5) Non-existent paths, but places where access is possible. For
example, a bike path passes close to the end of a cul-de-sac. There's
no actual paved or dirt path, but a cyclist could easily cross a metre
or two of grass (possibly dismounting). It seems crucial for routing
to make connections here. So I've been adding "highway=path". Is there
a better tag?

6) Places where a bike is probably permissible, but most people
wouldn't ride. (But I would :)) I'm not sure where the division of
responsibility for correctly handling bike routing lies, between the
OSM data, and the routing software. Is there any software smart enough
to give options like "how far are you willing to push the bike" or
"are you willing to cut across grass?" etc. An example is at a
university I used to ride through to get to work. I used to ride
around the side of an oval, and cut down through some trees on an a
true "unofficial path" - basically mountain biking. Do you mark it in
as an unofficial walking path, and tag it with appropriate mountain
biking paths, and assume the bike routing software is smart enough not
to route city bikes that way?

Maybe I'm looking for a distinction between "bicycle=no" and
"bicycle=forbidden".

7) Big open concrete spaces that are eminently navigable by
pedestrians and cyclists, but aren't exactly pedestrian malls. For
example, big spaces in business parks, or around big public buildings.
Mark them pedestrian anyway?

Thanks all,
Steve
Renaud MICHEL
2009-11-28 09:48:36 UTC
Permalink
Post by Steve Bennett
1) In the parks near me, there are lots of paths, which I guess were
probably intended for pedestrians, but cyclists use them too.
Sometimes paved, sometimes not. I've been tagging them "highway=path,
bicycle=yes" (to be safe).
If you use highway=path and not highway=footway, then you should also add
foot=yes (or even foot=designated if appropriate).
If I understand correctly
http://wiki.openstreetmap.org/wiki/Tag:highway=path
adding a bicycle=yes to a highway=path means that only bicycles are allowed
(whereas highway=path alone would mean "any non-motorized vehicle").
--
Renaud Michel
Konrad Skeri
2009-11-28 11:59:19 UTC
Permalink
highway=path
foot=yes
bicycle=no
mtb=yes

highway=footway implies foot=designated and highway=cycleway implies
bicycle=designated.
foot=yes means you can walk there while designated means it's the
primary choise of route for pedestrians.

See also http://wiki.openstreetmap.org/wiki/Key:smoothness

Konrad
Post by Steve Bennett
[...]
6) Places where a bike is probably permissible, but most people
wouldn't ride. (But I would :)) I'm not sure where the division of
responsibility for correctly handling bike routing lies, between the
OSM data, and the routing software. Is there any software smart enough
to give options like "how far are you willing to push the bike" or
"are you willing to cut across grass?" etc. An example is at a
university I used to ride through to get to work. I used to ride
around the side of an oval, and cut down through some trees on an a
true "unofficial path" - basically mountain biking. Do you mark it in
as an unofficial walking path, and tag it with appropriate mountain
biking paths, and assume the bike routing software is smart enough not
to route city bikes that way?
Maybe I'm looking for a distinction between "bicycle=no" and
"bicycle=forbidden".
[...]
Mike Harris
2009-11-28 12:20:29 UTC
Permalink
Steve

This is a big topic that has been very extensively discussed in this group
(and elsewhere). There is quite a range of opinion and, perhaps inevitably,
to some extent the opinions reflect (a) whether mappers see themselves
primarily as walkers, cyclists or ... mappers! and (b) the geographical
location of the mapper. The UK (or at least England and Wales) has developed
a quite sophisticated system based around the local legislation on public
rights of way - but, given your reference to Albert Park, you will probably
want to stand this on its head (:>). There are quite a lot of tags to look
at:

Highway=
Surface=
Tracktype=
Foot ? Bicycle ? Motorcar = yes ? permissive ? no
Designated =

I won?t bore you with my own practice (and this will perhaps avoid starting
up once more one of the long discussions we've had) beyond saying that I
would recommend that you avoid the use of highway=path except for very
ill-defined and unofficial paths (in your own words "an unpaved line of
footprints carved through the grass") and give preference to highway=footway
? track ? cycleway.

Given the controversies over the relative rights and priorities for
different classes of user (e.g. foot ? bicycle ? horse) and the large
regional differences between what is or is not permitted on different
classes of way (ranging from "everyman's right to wander" as in Germany and
most Nordic countries) to the strictly legalistic "public rights of way"
system in England where there is only a legal right where this is recorded
and defined) I would suggest that useful general guidelines are:

- record what is there on the ground by observation of state or signage.
- do not tag to make the maps render nicely - the renderers will eventually
catch up with what mappers do.
- add legal rights where you are sure about them e.g. by using the
designation= tag.
- be as explicit as possible as to what class of user may be able to use the
way (whether in practice or by right) as this will help clarify where one
person might call something a 'footway' and another a 'cycleway' - something
like foot=yes, bicycle=permissive is at least fairly explicit.

Before I get flamed - these are only my ideas and others may well differ -
but I've tried to keep it general as to practice and geography ...

Give my regards to Melbourne!

Mike Harris
-----Original Message-----
From: Steve Bennett [mailto:stevagewp at gmail.com]
Sent: 28 November 2009 08:24
To: talk at openstreetmap.org
Subject: [OSM-talk] Path vs footway vs cycleway vs...
Hi all,
(Apologies if this is the wrong list - still getting my
head around them all. Or this has been discussed extensively,
please point me at it)...
I'm doing a lot of mapping of pedestrian and bike paths
around my area, and am having trouble deciding when to use
path, when footway, and when cycleway. I'm particularly
troubled by the way Potlatch describes "path" as "unofficial
path" - making it sound like an unpaved line of footprints
carved through the grass.
1) In the parks near me, there are lots of paths, which I
guess were probably intended for pedestrians, but cyclists
use them too.
Sometimes paved, sometimes not. I've been tagging them
"highway=path, bicycle=yes" (to be safe).
2) Multi-use paths, like in new housing developments. Usually
paved, and connecting streets together.
3) Genuine multi-use paths along the sides of creeks or freeways.
Frequently with a dotted line down the middle. Most people
think of them as bike paths, but plenty of pedestrians use them too.
"highway=cycleway, foot=yes" seems the most satisfying, but
according to the definition, it should just be a "path"? I
tend to assume it's a cycleway if the gap between two
entrances ever exceeds a kilometre or so...
4) In Albert Park (home of the grand prix) near me, there are
lots of sealed paths that are wide enough for a car. They're
normally blocked off, and used mainly by contractors before
and after the grand prix.
The rest of the time, they're used by pedestrians and
cyclists. I had marked them "highway=unclassified" but now I
think "highway=track surface=paved" would be better?
5) Non-existent paths, but places where access is possible.
For example, a bike path passes close to the end of a
cul-de-sac. There's no actual paved or dirt path, but a
cyclist could easily cross a metre or two of grass (possibly
dismounting). It seems crucial for routing to make
connections here. So I've been adding "highway=path". Is
there a better tag?
6) Places where a bike is probably permissible, but most
people wouldn't ride. (But I would :)) I'm not sure where the
division of responsibility for correctly handling bike
routing lies, between the OSM data, and the routing software.
Is there any software smart enough to give options like "how
far are you willing to push the bike" or "are you willing to
cut across grass?" etc. An example is at a university I used
to ride through to get to work. I used to ride around the
side of an oval, and cut down through some trees on an a true
"unofficial path" - basically mountain biking. Do you mark it
in as an unofficial walking path, and tag it with appropriate
mountain biking paths, and assume the bike routing software
is smart enough not to route city bikes that way?
Maybe I'm looking for a distinction between "bicycle=no" and
"bicycle=forbidden".
7) Big open concrete spaces that are eminently navigable by
pedestrians and cyclists, but aren't exactly pedestrian
malls. For example, big spaces in business parks, or around
big public buildings.
Mark them pedestrian anyway?
Thanks all,
Steve
Steve Bennett
2009-11-28 13:26:06 UTC
Permalink
Thanks all, these are very good replies. I'll have to ponder for a
bit. One complication that I should perhaps have mentioned is at the
moment I'm doing a lot of the mapping based on NearMap aerial maps, so
I can't actually observe local practice to see what's going on. Which
is why I'm inferring as much as possible from things like the location
of the path: near houses, or in the middle of the bush... Sometimes
you can make out painted bike signs on the ground, sometimes you
can't.

Another tricky aspect is that the rules about what bikes can do vary
from council to council. It came up in the news recently that if you
ride a bike in a park in the City of Melbourne (ie, the most central
suburb), it's a $200 fine. No other inner city suburb bans bikes from
parks...

I'm still a bit confused by the notion of a "cycleway" - perhaps
because we don't use that term here at all, we say "bike path". OSM is
obviously an empirical process, and empirically, there is very little
or no difference between a "footpath" and a "bike path": they're both
paved, about a metre wide, and connect useful places together. In the
absence of signs, I don't see how there would be any satisfactory way
to decide whether something was a "cycleway" or a "footway", if those
are the only two choices. And with so little to distinguish them,
there must be a big grey area.

I guess I've seen true "cycleways" in places like the Netherlands,
where it's a genuine single-purpose path between two villages, crowded
with bikes. But there is barely anything like that here - it's always
multi-purpose. As an example:
http://www.vicroads.vic.gov.au/Home/RoadsAndProjects/RoadProjects/WesternSuburbs/DeerParkBypass.htm

Now, in common language, everyone would refer to this as a bike path.
It clearly has great interest to cyclists, as does the whole network
of "trails". But there's nothing about it that says it's a "bike path"
- it's called a "wellness trail" and is for "walking and cycling".
Instinctively, I want to tag it a cycleway...but there's absolutely
nothing to justify that. Nowhere will you see any primacy given to
cycling over walking. Conundrum.

Steve
Richard Fairhurst
2009-11-28 13:14:54 UTC
Permalink
Post by Steve Bennett
I'm doing a lot of mapping of pedestrian and bike paths around my
area, and am having trouble deciding when to use path, when footway,
and when cycleway. I'm particularly troubled by the way Potlatch
describes "path" as "unofficial path" - making it sound like an
unpaved line of footprints carved through the grass.
highway=footway -> a path intended for pedestrian use
highway=cycleway -> a path intended for pedestrian and cycle use
highway=bridleway -> a path intended for pedestrian and horse use[1]

Useful tags you can add to modify the above:

* "access" tags such as foot or bicycle. (So highway=cycleway, foot=no
would cover the rare case of a cycleway from which pedestrians are banned.)
* designation=whatever - for the official status of a path. (For
example, in the UK, you might have highway=bridleway,
designation=restricted_byway.)
* surface=tarmac | grass | dirt | gravel | whatever

highway=path is an invention of the wikifiddlers and not needed in 99%
of cases. The one case that isn't adequately covered by the above is
what some people call "pathways of desire" - informal shortcuts that
were never really laid out as a footpath. Like you say, an unpaved line
of footprints carved through the grass.
Post by Steve Bennett
1) In the parks near me, there are lots of paths, which I guess were
probably intended for pedestrians, but cyclists use them too.
highway=footway. You could add cycle=yes if bikes are permitted to use
them; or upgrade to highway=cycleway if they have the width/surface etc.
that characterises a cycleway.
Post by Steve Bennett
2) Multi-use paths, like in new housing developments. Usually paved,
and connecting streets together.
highway=cycleway.
Post by Steve Bennett
3) Genuine multi-use paths along the sides of creeks or freeways.
Frequently with a dotted line down the middle. Most people think of
them as bike paths, but plenty of pedestrians use them too.
highway=cycleway. If there's a dotted line you could add segregated=yes.
Post by Steve Bennett
4) In Albert Park (home of the grand prix) near me, there are lots of
sealed paths that are wide enough for a car. They're normally blocked
off, and used mainly by contractors before and after the grand prix.
The rest of the time, they're used by pedestrians and cyclists. I had
marked them "highway=unclassified" but now I think "highway=track
surface=paved" would be better?
Without knowing the exact place, probably something like:
highway=service, access=private, bicycle=permissive, foot=permissive
Post by Steve Bennett
5) Non-existent paths, but places where access is possible. For
example, a bike path passes close to the end of a cul-de-sac. There's
no actual paved or dirt path, but a cyclist could easily cross a metre
or two of grass (possibly dismounting). It seems crucial for routing
to make connections here. So I've been adding "highway=path". Is there
a better tag?
highway=path is well-suited for this.
Post by Steve Bennett
6) Places where a bike is probably permissible, but most people
wouldn't ride. (But I would :)) I'm not sure where the division of
responsibility for correctly handling bike routing lies, between the
OSM data, and the routing software. Is there any software smart enough
to give options like "how far are you willing to push the bike" or
"are you willing to cut across grass?" etc.
cyclestreets.net is an OSM-based routing site with an option for pushing
your bike, so yes, there is.
Post by Steve Bennett
7) Big open concrete spaces that are eminently navigable by
pedestrians and cyclists, but aren't exactly pedestrian malls.
I have no idea about landuse types so will leave this to others!


All IMO, of course. I've cross-posted this to the tagging@ list which is
better suited for this kind of discussion.

cheers
Richard
Steve Bennett
2009-11-28 13:37:12 UTC
Permalink
On Sun, Nov 29, 2009 at 12:14 AM, Richard Fairhurst
Post by Richard Fairhurst
highway=footway -> a path intended for pedestrian use
highway=cycleway -> a path intended for pedestrian and cycle use
highway=bridleway -> a path intended for pedestrian and horse use[1]
Boy, I like this way of thinking. Of course, it must be controversial
given the preceding comments, but it does make a lot of sense.

Not really sure what a bridleway is in practice, but we do have rail
trails that allow all three modes, and a couple of long distance
trails that allow all three, but are really best suited to horses (too
far between camps for walkers, too rough for cyclists).
Post by Richard Fairhurst
* "access" tags such as foot or bicycle. (So highway=cycleway, foot=no
would cover the rare case of a cycleway from which pedestrians are banned.)

I've used this a few times. It crops up in my area where there are two
distinct paths, one for bikes and one for pedestrians, and they follow
slightly different routes. (See the Bay Trail between St Kilda and
Elwood, Victoria, Australia for example...)

To expand on the semantics of what you posted:

highway=footway -> purpose built path for pedestrians
highway=cycleway -> purpose built path for pedestrians and/or
cyclists, with all the characteristics of a bike path (no steps, no
kerbs, width >1m), no restrictions against bikes

Agree? Then we can keep it totally empirical and objective, without
worrying about whether the thing is labelled "xxx bike path" or was
intended for that purpose. In particular, I'm thinking of lots of
paths that were built with pedestrians in mind, before the cycling
revolution came along...
Post by Richard Fairhurst
highway=cycleway doesn't mean cycles have priority. It just means it's intended for pedestrian and cycle use. There's no suggestion of primacy for either.
Cool. So again, "cycleway" is a statement of the quality and
attributes of the path, rather than implying any design decisions,
rules, usage etc.

Next question: how popular is this viewpoint? Is this a minority way
of thinking?

Steve
Ben Laenen
2009-11-28 14:09:19 UTC
Permalink
Post by Steve Bennett
On Sun, Nov 29, 2009 at 12:14 AM, Richard Fairhurst
Post by Richard Fairhurst
highway=footway -> a path intended for pedestrian use
highway=cycleway -> a path intended for pedestrian and cycle use
highway=bridleway -> a path intended for pedestrian and horse use[1]
Boy, I like this way of thinking. Of course, it must be controversial
given the preceding comments, but it does make a lot of sense.
And at one time it was that easy in OSM, but the real world really isn't. In
some countries it may work fine, but in other countries the distinction
between the three has no connection with the actual situation and would
introduce a number of ambiguities where you don't really know anymore whether
something is allowed or not.

Take cycleways for example. Over here mopeds are allowed on paths that are
signed as cycleway. Now, on the other hand we also had paths which weren't
cycleways but allowed bicycles (but no mopeds) tagged as cycleway. Conflict
between the two: would a route planner now allow mopeds on them or not? Sure,
one could explicitly tag the moped=yes/no but (a) mappers forget about it, and
(b) even if they don't, they often do not know the exact rules. And not
forgetting that (c) traffic code isn't some static thing, it changes over time
and what has been allowed on a certain path with certain signs, may not be in
future.

Hence the addition of highway=path was actually a welcome additional tag. Now
we can tag the paths that are legal cycleways as highway=cycleway (and
likewise for footpaths and bridleways), and other paths with the generic
highway=path. The traffic signs on those paths can then be translated to
access tags.

Greetings
Ben
Steve Bennett
2009-11-28 14:24:57 UTC
Permalink
Post by Ben Laenen
And at one time it was that easy in OSM, but the real world really isn't. In
some countries it may work fine, but in other countries the distinction
between the three has no connection with the actual situation and would
introduce a number of ambiguities where you don't really know anymore whether
something is allowed or not.
Ok, since I'm new here, I'll ask the obvious question: does it matter
whether this stuff is done the same across different countries? Is it
not ok if "cycleway" has slightly different semantics in different
jurisdictions?

(Australian bias showing, I'm unable to conceive of the idea of
cycling from one country to another...)
Richard Weait
2009-11-28 19:42:52 UTC
Permalink
Post by Steve Bennett
Ok, since I'm new here,
You're new here? Welcome to OSM.
Post by Steve Bennett
I'll ask the obvious question: does it matter
whether this stuff is done the same across different countries? Is it
not ok if "cycleway" has slightly different semantics in different
jurisdictions?
A map is an abstraction and can not hope to perfectly represent all of
the wonderful variations of 'things' we see. There are likely to be
several ways to do some of the things that you want to do. Some of
these variations will have subtle benefits and some will be matters of
personal preference. Others will be noticeably different than what you
will see in other jurisdictions.

Look to see what other are doing locally and in similar places.
Learn and adapt what you see as best practice in other places.
Have fun.
Mike Harris
2009-11-29 10:24:51 UTC
Permalink
Good advice ... +1!

Mike Harris
-----Original Message-----
From: Richard Weait [mailto:richard at weait.com]
Sent: 28 November 2009 19:43
To: Steve Bennett
Cc: talk at openstreetmap.org; tagging at openstreetmap.org
Subject: Re: [OSM-talk] Path vs footway vs cycleway vs...
On Sat, Nov 28, 2009 at 9:24 AM, Steve Bennett
Post by Steve Bennett
Ok, since I'm new here,
You're new here? Welcome to OSM.
Post by Steve Bennett
I'll ask the obvious question: does it matter whether this stuff is
done the same across different countries? Is it not ok if
"cycleway"
Post by Steve Bennett
has slightly different semantics in different jurisdictions?
A map is an abstraction and can not hope to perfectly
represent all of the wonderful variations of 'things' we see.
There are likely to be several ways to do some of the things
that you want to do. Some of these variations will have
subtle benefits and some will be matters of personal
preference. Others will be noticeably different than what you
will see in other jurisdictions.
Look to see what other are doing locally and in similar places.
Learn and adapt what you see as best practice in other places.
Have fun.
Lesi
2009-11-28 14:28:46 UTC
Permalink
The footway/cycleway/path choas is the one of the biggest drawbacks of OSM.

Here's my approach:
- A footway is a mostly paved way in a city. It's a way which was mostly
built by an authority. You can walk on it safely in high heels.
- A path is a narrow way, which is mostly not paved and was not built by
somebody. This can be short cuts in cities, ways in a forest which are to
narrow to be tagged as tracks or hiking trails in the mountains. If it's
raining you could get dirty shoes.
You can indicate that the path is (not) suitable for bikes with
bicycle=yes/no.
You can ride with your bike everywhere in my area, so I do not use cycleway.

lesi
Mike Harris
2009-11-29 10:24:51 UTC
Permalink
Not to suggest that there is a 'right' or a 'wrong' approach - but merely to
note that I (England mostly) - and I believe some others in England and
perhaps elsewhere) have a different approach - this is, I stress, what I
currently do - and has evolved as a result of my own (limited) experience in
mapping and participation in various group discussions:

1. All ways that are not available other than to pedestrians are
highway=footway - whether urban paved footways or rural unpaved 'footpaths'.
Even a rural 'footpath' that is barely discernible where it crosses, for
example, pasture, is highway=footway if it is a legal public footpath.

2. Highway=path is only used for a route - usually ill-defined and often in
upland areas where the precise legal line of a public footpath is often less
meaningful than the customary route (e.g. up a mountain) - in the sense that
people walk it.

3. Highway=track is used similarly for something that is wider and, at least
in principle, available for use by a four-wheeled (off-road e.g. a farm
tractor) vehicle.

3. I would then define legal status, where known, using a designated= tag
and surface condition using a combination of tracktype= and/or surface= as
appropriate. I would also add ref= where the reference number of the way was
known.

4. I would always add foot=yes (or at least foot=permissive) for clarity and
also add bicycle ? horse = yes ? permissive ? no as appropriate.

5. I would reserve highway=cycleway for something that was (a) built
primarily for use by bicycles - whether beside a motor road or not and was
(b) (only relevant in England and Wales) not a public
footpath/bridleway/byway (as these have legally defined rights for different
classes of user). I would then add foot=yes (unless pedestrians were
actually forbidden) for additional clarity and perhaps an indication as to
whether it was a shared way for cyclists and walkers or a longitudinally
divided dual use way.

6. I would use a route relation to define medium- / long-distance routes -
e.g. a long-distance path or a national/regional cycleway - adding names and
reference numbers to the relation.

Again, I stress, this is just what I do - in the interest of transparency -
and not in any way to suggest that it is better or worse than what Lesi or
anyone else has adopted. This is OSM - the ultimate popular democracy!

Have fun mapping!

Mike Harris
-----Original Message-----
From: Lesi [mailto:lesi at lesi.is-a-geek.net]
Sent: 28 November 2009 14:29
To: talk at openstreetmap.org
Subject: Re: [OSM-talk] Path vs footway vs cycleway vs...
The footway/cycleway/path choas is the one of the biggest
drawbacks of OSM.
- A footway is a mostly paved way in a city. It's a way which
was mostly built by an authority. You can walk on it safely
in high heels.
- A path is a narrow way, which is mostly not paved and was
not built by somebody. This can be short cuts in cities, ways
in a forest which are to narrow to be tagged as tracks or
hiking trails in the mountains. If it's raining you could get
dirty shoes.
You can indicate that the path is (not) suitable for bikes
with bicycle=yes/no.
You can ride with your bike everywhere in my area, so I do
not use cycleway.
lesi
Cartinus
2009-11-28 20:33:41 UTC
Permalink
Post by Steve Bennett
Next question: how popular is this viewpoint? Is this a minority way
of thinking?
It was the only viewpoint before highway=path was "invented". Now it is one of
several competing viewpoints without a clear winner.
--
m.v.g.,
Cartinus
Richard Fairhurst
2009-11-28 13:30:34 UTC
Permalink
Post by Steve Bennett
Instinctively, I want to tag it a cycleway...but there's absolutely
nothing to justify that. Nowhere will you see any primacy given to
cycling over walking. Conundrum.
highway=cycleway doesn't mean cycles have priority. It just means it's
intended for pedestrian and cycle use. There's no suggestion of primacy
for either.

(Incidentally, I missed out the footnote from my last mail, which was
going to say that in some countries (like the UK) cycles are permitted
on bridleways; nonetheless it's most sensible to treat highway=bridleway
as a path for pedestrian and horse use, and tag over and above that if
it's a cyclable one.)

cheers
Richard
Lester Caine
2009-11-29 08:23:06 UTC
Permalink
Post by Richard Fairhurst
Post by Steve Bennett
Instinctively, I want to tag it a cycleway...but there's absolutely
nothing to justify that. Nowhere will you see any primacy given to
cycling over walking. Conundrum.
highway=cycleway doesn't mean cycles have priority. It just means it's
intended for pedestrian and cycle use. There's no suggestion of primacy
for either.
(Incidentally, I missed out the footnote from my last mail, which was
going to say that in some countries (like the UK) cycles are permitted
on bridleways; nonetheless it's most sensible to treat highway=bridleway
as a path for pedestrian and horse use, and tag over and above that if
it's a cyclable one.)
http://www.horsedata.co.uk/Jargon.htm I think nicely covers most of the options.
But 'cycleway' is still a rather woolly term, with many specially constructed
cycle routes using that term only for a fully-segregated cycle route, however
sustrans.org.uk works on the basis that the national cycle network is also for
'walkers, wheelchair uses and horseriders' and this is where the 'path'
designation sort of came from since many vehicle free routes are not footway,
cycleway or bridalway.

Personally I would only use cycleway where the 'path' was specifically
restricted to cycles for reasons of safety - such as the segregated routes that
form part of a main vehicle way, lanes on the side such as described in
http://www.cycling.bham.ac.uk/other/RoadDesignTerminology.shtml but at this
point we are getting back to the 'micromapping' question on when does a single
'way' get replaced with a group of parallel ways each with their correct
designation ..... around here it is the footpaths which do not follow directly
the 'roadway' and many of them are 'pavements' where cycling is specifically banned.
--
Lester Caine - G8HFL
-----------------------------
Contact - http://lsces.co.uk/wiki/?page=contact
L.S.Caine Electronic Services - http://lsces.co.uk
EnquirySolve - http://enquirysolve.com/
Model Engineers Digital Workshop - http://medw.co.uk//
Firebird - http://www.firebirdsql.org/index.php
John F. Eldredge
2009-11-28 14:35:29 UTC
Permalink
Underwater bicycling, the next Olympic sport...

-------Original Email-------
Subject :Re: [OSM-talk] Path vs footway vs cycleway vs...
From :stevagewp at gmail.com
Date :Sat Nov 28 08:24:57 America/Chicago 2009

(Australian bias showing, I'm unable to conceive of the idea of
cycling from one country to another...)
--
John F. Eldredge -- john at jfeldredge.com
"Reserve your right to think, for even to think wrongly is better than not to think at all." -- Hypatia of Alexandria
Roy Wallace
2009-11-28 23:01:18 UTC
Permalink
I have a couple of thoughts:

1) Re: connecting paths across small grass areas - don't mark a path
where there isn't one, and especially don't do it for the purpose of
trying to make routers work better. Map reality - that will always
work best in the long term. (just my personal preference)

2) Re: when to use path/footway/cycleway etc. - firstly, I prefer
highway=path because it is more extensible. Any
highway=footway/cycleway/bridleway can be expressed in terms of a
highway=path with additional access tags. In this way, using
highway=path can be more explicit, because of ongoing disagreements in
the definition of footway/cycleway/bridleway.

3) Re: what does <TAG> really mean? - rather than everyone giving
their personal opinion on e.g. what highway=path means, for new users
I would strongly recommend reading the wiki carefully and using that.
I'm sure there are plenty of mappers who read the wiki and nothing
else, and if consistency is the goal, I think the wiki should serve to
document the current consensus as well as current disagreements. Of
course, the wiki needs improving, and I personally think we should
make this a priority. See, for example, some of the latest efforts to
improve the situation:
http://wiki.openstreetmap.org/wiki/Consolidation_footway_cycleway_path
Liz
2009-11-28 23:21:36 UTC
Permalink
Post by Roy Wallace
I would strongly recommend reading the wiki carefully and using that.
but Roy, the wiki is written by committee and it is a good example of the
failure of the committee process
the minority report cannot be distinguished from the majority report

so a newbie reading the wiki is just going to become confused when it is a
non-vehicular way
Roy Wallace
2009-11-28 23:49:44 UTC
Permalink
Post by Liz
Post by Roy Wallace
I would strongly recommend reading the wiki carefully and using that.
but Roy, the wiki is written by committee and it is a good example of the
failure of the committee process
the minority report cannot be distinguished from the majority report
so a newbie reading the wiki is just going to become confused when it is a
non-vehicular way
I think you missed my point - let me clarify.

If a newbie asks "hey guys, what's a footway?" and they get 50
responses saying "well, I think it's..." and "well I've been using..."
and "no, no, it's really...", that will get us nowhere. Plus, what
about the newbies who *don't ask?!*

The newbie reading these conflicting responses either 1) becomes
confused, or 2) begins to think that best practice is to invent your
own meaning for existing tags and then pass this secret knowledge on
to only the newbies who ask via email. This is not a good outcome.

Please let me stress that I am not saying the wiki is in a good state!
But it is the best thing to refer to as a reference for tag meanings,
because it is *documented*. That is, for the 10,000's of mapper who
are out there adding footways right now and are *not on this list*,
one must assume they are doing so on the basis of the definition in
the wiki. That is certainly what I did and will continue to do.

So if consistency is the goal, you cannot rely on various personal
opinions that exist only in people's minds and in email discussions
from time to time (which no doubt only a small proportion of mappers
ever read). You must write it down for reference. And if what's
written down has flaws, they must be fixed.

Note also that by the wiki serving as a "reference" I do not mean that
the wiki page for, say, footway must give only the one "true"
definition. It should 1) document the usage of tags as they occur in
the database, 2) detail any ongoing controversy and 3) if a consensus
exists, give a clear recommendation on how the tag should be used by
new mappers.
Nop
2009-11-29 00:34:19 UTC
Permalink
Hi!
Post by Roy Wallace
The newbie reading these conflicting responses either 1) becomes
confused, or 2) begins to think that best practice is to invent your
own meaning for existing tags and then pass this secret knowledge on
to only the newbies who ask via email. This is not a good outcome.
The newbie - who usually assumes that there is a simple and
straightforward answer to the simple question "how to I tag a footway" -
becomes confused - and frustrated that such a basic thing is unsolved
and not looking like it's going to be solved one of these years. To the
newcomer, this is somewhere between unexpected and crazy.
Post by Roy Wallace
So if consistency is the goal, you cannot rely on various personal
opinions that exist only in people's minds and in email discussions
from time to time (which no doubt only a small proportion of mappers
ever read). You must write it down for reference. And if what's
written down has flaws, they must be fixed.
No help there. The major contractiory interpretations of the tags around
this topic are all "documented" in the wiki in contradictory ways. It
just depends on which page you find first and what conlusions you derive
from rather fuzzy definitions.
Post by Roy Wallace
Note also that by the wiki serving as a "reference" I do not mean that
the wiki page for, say, footway must give only the one "true"
definition. It should 1) document the usage of tags as they occur in
the database, 2) detail any ongoing controversy and 3) if a consensus
exists, give a clear recommendation on how the tag should be used by
new mappers.
1) The same tags are used with up to 5 different meanings - usually one
wiki page only states one interpretation, but there are many different
pages.
2) AFAIK the only attempt at a neutral display of the different opinions
is here:
http://wiki.openstreetmap.org/wiki/Consolidation_footway_cycleway_path
3) There has never been anything approaching a consensus. Not even
close. The discussion has been going around in circles since I first
thought there had to be a simple answer to a simple question. Which is
about a year. :-)

bye
Nop
Roy Wallace
2009-11-29 01:20:27 UTC
Permalink
Post by Nop
Post by Roy Wallace
So if consistency is the goal, you cannot rely on various personal
opinions that exist only in people's minds and in email discussions
from time to time (which no doubt only a small proportion of mappers
ever read). You must write it down for reference. And if what's
written down has flaws, they must be fixed.
No help there. The major contractiory interpretations of the tags around
this topic are all "documented" in the wiki in contradictory ways. It just
depends on which page you find first and what conlusions you derive from
rather fuzzy definitions.
I know. I didn't mean to say the *content* of the wiki is necessarily
good, just that I think the *concept* of the wiki is a better way to
aim for consistency than throwing around personal opinions from time
to time.
Anthony
2009-11-29 00:33:13 UTC
Permalink
Post by Roy Wallace
1) Re: connecting paths across small grass areas - don't mark a path
where there isn't one, and especially don't do it for the purpose of
trying to make routers work better. Map reality - that will always
work best in the long term. (just my personal preference)
When is there a path and when is there not a path? I walk through an
area of grass every time I go to the park near my house. Isn't that a
"path" which is part of "reality"?
Post by Roy Wallace
3) Re: what does <TAG> really mean? - rather than everyone giving
their personal opinion on e.g. what highway=path means, for new users
I would strongly recommend reading the wiki carefully and using that.
"A generic path, either multi-use, or unspecified usage."

Umm, okay. I take that to mean anything course of travel that isn't
covered by one of the other highway tags.
Roy Wallace
2009-11-29 01:15:14 UTC
Permalink
When is there a path and when is there not a path? ?I walk through an
area of grass every time I go to the park near my house. ?Isn't that a
"path" which is part of "reality"?
An area of grass is - to me - not a path. A path, IMHO, is something
that exists independently of people walking or not walking on it (i.e.
usually you can *see* that it resembles a path).
Post by Roy Wallace
3) Re: what does <TAG> really mean? - rather than everyone giving
their personal opinion on e.g. what highway=path means, for new users
I would strongly recommend reading the wiki carefully and using that.
"A generic path, either multi-use, or unspecified usage."
Umm, okay. ?I take that to mean anything course of travel that isn't
covered by one of the other highway tags.
That's fair enough. My main point was that you can at least be assured
that other mappers are using the same documentation (the wiki as a
whole) to decide how to tag their ways. If you ask on this email list,
you cannot be assured of that.

As an aside, about highway=path...the definition of "generic" is
"descriptive of all members of a genus", so I take it to mean that -
quite intuitively - all paths are a kind of path, regardless of
whether you can ride a bicycle or walk or snowmobile on them :) But
I'm not going to get into this discussion again - instead let's go and
improve http://wiki.openstreetmap.org/wiki/Consolidation_footway_cycleway_path
Anthony
2009-11-29 02:20:31 UTC
Permalink
Post by Roy Wallace
When is there a path and when is there not a path? ?I walk through an
area of grass every time I go to the park near my house. ?Isn't that a
"path" which is part of "reality"?
An area of grass is - to me - not a path.
Never? Or just not generally?

What if the grass is slightly bare?
Loading Image...

Cut short? Loading Image...
Loading Image...

Through an otherwise impassible area?
Loading Image...

Marked by a sign? Loading Image...
Post by Roy Wallace
A path, IMHO, is something
that exists independently of people walking or not walking on it (i.e.
usually you can *see* that it resembles a path).
Usually, or always?

Usually, fine, I agree. Always, that just doesn't coincide with my
definition of "path". To me, the fact that you can usually recognize
a path is an effect, not a cause.

If there were some other tag for me to use (say highway=grass), fine.
But none of the other highway tags are appropriate, and the routing
information needs to be designated somehow. The area of grass I have
in mind exists in a legal right of way. It's not like I'm talking
about cutting through someone's backyard. It's a perfectly legitimate
path of travel. It should provided in walking directions. And that
means having some sort of highway tag.
Roy Wallace
2009-11-29 03:15:53 UTC
Permalink
Post by Roy Wallace
An area of grass is - to me - not a path.
Never? ?Or just not generally?
I'll rephrase. The following, IMHO, are not sufficient reasons to tag
an area of grass as a path: 1) you walk on it; 2) you think it would
help routing. Analogy: 1) Just because you sit on something, that
doesn't make it a chair; 2) Just because you want others to be
recommended to sit on it, that doesn't make it a chair.

The only reason I would tag an area of grass as a path is if, when I
asked a typical stranger, "hey, is that over there a path?", they
replied yes. If I ask "is this a chair?"...you get the picture.

In that sense, of course, the photos you linked to are paths. Common sense.
Post by Roy Wallace
A path, IMHO, is something
that exists independently of people walking or not walking on it (i.e.
usually you can *see* that it resembles a path).
Usually, or always?
Um... so the question is, if you can't see a path, can it still be a
path? Answer: No, because otherwise your mapping is not verifiable:
http://wiki.openstreetmap.org/wiki/Verifiability.
If there were some other tag for me to use (say highway=grass), fine.
But none of the other highway tags are appropriate, and the routing
information needs to be designated somehow. ?The area of grass I have
in mind exists in a legal right of way. ?It's not like I'm talking
about cutting through someone's backyard. ?It's a perfectly legitimate
path of travel. ?It should provided in walking directions. ?And that
means having some sort of highway tag.
I don't have an easy answer for your problem. I would urge caution,
though, in tagging things that aren't verifiable.

Actually, I remember trekking recently, using an OSM map, that
connected one track to another. The tracks actually *weren't
connected* in any way other than through a short stint through dense
forest. This is the problem: when you tag in order to have things
"provided in walking directions", this can lead you astray. Oh, and if
you like highway=grass, use that!
Anthony
2009-11-29 04:29:41 UTC
Permalink
Post by Roy Wallace
The following, IMHO, are not sufficient reasons to tag
an area of grass as a path: 1) you walk on it; 2) you think it would
help routing. Analogy: 1) Just because you sit on something, that
doesn't make it a chair; 2) Just because you want others to be
recommended to sit on it, that doesn't make it a chair.
Bad analogy. If I look in a dictionary under "chair", there is no
definition which says "a thing that is sat upon". But if I look under
"path", there is a definition which says "a route, course, or track
along which something moves".
Post by Roy Wallace
Post by Anthony
Post by Roy Wallace
A path, IMHO, is something
that exists independently of people walking or not walking on it (i.e.
usually you can *see* that it resembles a path).
Usually, or always?
Um... so the question is, if you can't see a path, can it still be a
path?
No, my question was whether you really meant to use the word "usually".
Post by Roy Wallace
http://wiki.openstreetmap.org/wiki/Verifiability.
The fact that an area of land is within a legally defined right of way
is verifiable. The fact that it is suitable for travel is verifiable.
The fact that people use it for travel is verifiable.

I suppose in that sense I can *see* that it resembles a path.
Post by Roy Wallace
Oh, and if you like highway=grass, use that!
I like highway=path. More general.
Anthony
2009-11-29 04:48:29 UTC
Permalink
Post by Anthony
Post by Roy Wallace
Oh, and if you like highway=grass, use that!
I like highway=path.
With surface=grass, of course!
Mike Harris
2009-11-29 14:13:06 UTC
Permalink
Btw - no need for highway=grass, why not use highway=path (or =footway, see
previous message) + surface=grass (which seems well-established).

Mike Harris
-----Original Message-----
From: Anthony [mailto:osm at inbox.org]
Sent: 29 November 2009 04:30
To: Roy Wallace
Cc: talk at openstreetmap.org
Subject: Re: [OSM-talk] Path vs footway vs cycleway vs...
On Sat, Nov 28, 2009 at 10:15 PM, Roy Wallace
The following, IMHO, are not sufficient reasons to tag an area of
grass as a path: 1) you walk on it; 2) you think it would help
routing. Analogy: 1) Just because you sit on something,
that doesn't
make it a chair; 2) Just because you want others to be
recommended to
sit on it, that doesn't make it a chair.
Bad analogy. If I look in a dictionary under "chair", there
is no definition which says "a thing that is sat upon". But
if I look under "path", there is a definition which says "a
route, course, or track along which something moves".
Post by Anthony
Post by Roy Wallace
A path, IMHO, is something
that exists independently of people walking or not
walking on it (i.e.
Post by Anthony
Post by Roy Wallace
usually you can *see* that it resembles a path).
Usually, or always?
Um... so the question is, if you can't see a path, can it
still be a
path?
No, my question was whether you really meant to use the word
"usually".
http://wiki.openstreetmap.org/wiki/Verifiability.
The fact that an area of land is within a legally defined
right of way is verifiable. The fact that it is suitable for
travel is verifiable.
The fact that people use it for travel is verifiable.
I suppose in that sense I can *see* that it resembles a path.
Oh, and if you like highway=grass, use that!
I like highway=path. More general.
Anthony
2009-11-29 14:49:33 UTC
Permalink
Post by Mike Harris
Btw - no need for highway=grass, why not use highway=path (or =footway, see
previous message) + surface=grass (which seems well-established).
I was just proposing a compromise. I don't care what the tags are so
long as they are well-defined. highway=qwijibo (or
highway=invisible_path) is fine with me.

The wiki right now says that highway=path is to be used for paths,
which is incredibly unhelpful. There are also some examples, which
suggest to me that a "path", as used by OSM, means essentially "any
highway (place open to the public where people travel) which doesn't
fall under another highway=* tag".

If I thought the wiki was a productive work environment, I'd try to
add that more specific definition there. But I don't.

If that isn't the definition, then I propose "highway=highway", to
have that definition. Personally, I don't see much sense
distinguishing between different types of "highway"s except in areas
where there is a formal legal designation. Number of lanes should be
represented by lanes=*. Maximum speed should be represented by
maxspeed=*. Surfaces can be described with surface=*. Access is
determined by access=*. Importance can then be determined,
objectively, during a preprocessing stage which factors in all these
conditions along with the physical connections. It should then be the
job of computers to combine all those elements together and decide
what colors to paint things.

I don't see that happening, so I'll just make my best guess as to
which highway=* tag to use, and not particularly worry about it
(except when someone tells me that I can't use *any* of the highway=*
tags for something which ought to be in the routing network).
Cartinus
2009-11-29 02:23:16 UTC
Permalink
Post by Nop
2) AFAIK the only attempt at a neutral display of the different opinions
http://wiki.openstreetmap.org/wiki/Consolidation_footway_cycleway_path
That page is far from neutral, because the only solutions it offers are doing
something with the path tag.
Post by Nop
That's fair enough. My main point was that you can at least be assured
that other mappers are using the same documentation (the wiki as a
whole) to decide how to tag their ways. If you ask on this email list,
you cannot be assured of that.
Actually you can't, because there is a whole horde of experienced mappers that
gave up on the wiki-mess. But they do speak up from time to time on the
mailinglists.
--
m.v.g.,
Cartinus
Nop
2009-11-29 08:31:27 UTC
Permalink
Hi!
Post by Cartinus
Post by Nop
2) AFAIK the only attempt at a neutral display of the different opinions
http://wiki.openstreetmap.org/wiki/Consolidation_footway_cycleway_path
That page is far from neutral, because the only solutions it offers are doing
something with the path tag.
It is an attempt. If you find something missing or have another
suggestion for a solution, why don't you add it?

bye
Nop
Cartinus
2009-11-29 17:34:27 UTC
Permalink
Post by Nop
It is an attempt. If you find something missing or have another
suggestion for a solution, why don't you add it?
Because I am not allowed to.

The page starts with stating that if you don't agree with the problem, then
you are not allowed to contribute.

Then it tells us about all the problems that the "old" fuzzy definitions
cause. Next it barely recognises that the path tag isn't perfect either.

I am of the opinion that the "old" fuzzy definitions weren't a problem at all
and the path tag should only be used for things that really don't fit in
them. (Like the snowmobile trail.)
--
m.v.g.,
Cartinus
Cartinus
2009-11-29 19:19:28 UTC
Permalink
Hi!
Post by Cartinus
I am of the opinion that the "old" fuzzy definitions weren't a problem at
all and the path tag should only be used for things that really don't fit
in them. (Like the snowmobile trail.)
I guess you are right. Adding a sixth contradictory opinion probably
will not help.
If you negate the existence of a problem that has been widely confirmed,
you're not likely to contribute to a solution.
bye
Nop
Except that I am far from alone with my opinion. See e.g. Richards explanation
somewhere at the start of this thread and the widespread opposition the path
tag gets.
--
m.v.g.,
Cartinus

P.S. Please keep list discussions on the list.
Nop
2009-11-29 21:53:58 UTC
Permalink
Hi!
Post by Cartinus
If you negate the existence of a problem that has been widely confirmed,
you're not likely to contribute to a solution.
Except that I am far from alone with my opinion. See e.g. Richards explanation
somewhere at the start of this thread and the widespread opposition the path
tag gets.
EVERY contradictory interpretation has a substantial number of followers
- that IS the problem. Richards view works only in the UK and fails
terribly in Germany and other countries. But sorry, I really am fed up
with the pointless discussions on this matter, so I'll refrain from
plucking apart the details. It has all been said before.

bye
Nop
Steve Bennett
2009-11-29 22:10:15 UTC
Permalink
Post by Nop
EVERY contradictory interpretation has a substantial number of followers
- that IS the problem. Richards view works only in the UK and fails
terribly in Germany and other countries. But sorry, I really am fed up
with the pointless discussions on this matter, so I'll refrain from
plucking apart the details. It has all been said before.
Before you go, do you think there is potential at least to have
consistency within each country?

Steve
Cartinus
2009-11-30 00:30:48 UTC
Permalink
Post by Steve Bennett
Before you go, do you think there is potential at least to have
consistency within each country?
I'm not the one that leaves, but the answer would be yes.

It's fairly simple to put foot=no on all cycleways in what is probably the
only country with rules for cycleways that are so strict.

The often mentioned German paths with a white line in the middle (that
separates cyclists and pedestrians) could have been done with
highway=cycleway+footway=lane or something similar. That is analogous to how
we treat e.g. a tertiary road with cycle lanes.

etc. etc. etc.

The path crowd however wanted "one solution for everything" and can't accept
that people didn't want to redo all existing tagging. Especially not in
places where it simply works.

The result is that some people use path as it is designed, some people don't
use path at all and other people use path for what the translated word path
means in their language (often some kind of unpaved footway).
--
m.v.g.,
Cartinus
Richard Fairhurst
2009-11-30 00:38:07 UTC
Permalink
Post by Cartinus
It's fairly simple to put foot=no on all cycleways in what is probably
the only country with rules for cycleways that are so strict.
Indeed.
Post by Cartinus
The often mentioned German paths with a white line in the middle
(that separates cyclists and pedestrians) could have been done with
highway=cycleway+footway=lane or something similar.
highway=cycleway, segregated=yes . There are zillions of these in Britain
too.

(I'd also observe that it's already accepted that you parse highway values
according to the country. A UK primary road is not the same at all as a
French voie express, but both of them are tagged with highway=trunk.)

cheers
Richard
--
View this message in context: http://old.nabble.com/Path-vs-footway-vs-cycleway-vs...-tp26551214p26568023.html
Sent from the OpenStreetMap - General mailing list archive at Nabble.com.
Steve Bennett
2009-11-30 01:33:00 UTC
Permalink
On Mon, Nov 30, 2009 at 11:38 AM, Richard Fairhurst
Post by Richard Fairhurst
Post by Cartinus
It's fairly simple to put foot=no on all cycleways in what is probably
the only country with rules for cycleways that are so strict.
Indeed.
Yeah, but from the point of view of a resident of that country, doing
the mapping..."why should I put 'foot=no' on every cycleway? that's so
redundant!"

I have a similar issue with the suggestion that I'm supposed to mark
every mini_roundabout "direction=clockwise". I refuse. Some day the
renderers and routers will get smart enough to figure out that EVERY
mini_roundabout goes clockwise in this country, and every other
left-drive country.

So do we just need a more managed approach to interjurisdictional
variation? We have some ad hoc tables for things like freeways. Why
not make this approach more structured, and possibly encoded, so that
we can use plain old "cycleway" in different countries, and have a
table that explains to routers what that means?

Currently the definitions on each page are very vague, as they try and
capture commonalities across all countries, even though actual
practice is more specific.

Steve
Mike Harris
2009-11-30 09:23:27 UTC
Permalink
As an Englander who has lived, albeit briefly, in Germany I do perhaps
recognise the difference between Germany and England as regards cycleways. I
think - but am not certain - that Germany is relatively unusual in having a
lot of cycleways that are NOT for pedestrians (foot=no) as Cartinus
suggests.

However, segregated cycleways are - I believe - common in both countries
(and others) - i.e. there are parallel 'lanes' for cyclists and pedestrians
(even if the separation / segregation is only by a painted white line - and
[only in England, of course, never in Germany (;>)] - often ignored by both
classes of user). Rather than use something a bit complicated like
"highway=cycleway+footway=lane" I tend to prefer the advice given in the
wiki at:

http://wiki.openstreetmap.org/wiki/Tag:access%3Ddesignated

which even addresses the dreaded snowmobile issue.

In a more general vein the use of the designated= tag has 'solved' a number
of related problems - at least for me.

But long live chaos, anarchy and OSM ... (:>)

Mike Harris
-----Original Message-----
From: Cartinus [mailto:cartinus at xs4all.nl]
Sent: 30 November 2009 00:31
To: talk at openstreetmap.org
Subject: Re: [OSM-talk] Path vs footway vs cycleway vs...
Post by Steve Bennett
Before you go, do you think there is potential at least to have
consistency within each country?
I'm not the one that leaves, but the answer would be yes.
It's fairly simple to put foot=no on all cycleways in what is
probably the only country with rules for cycleways that are so strict.
The often mentioned German paths with a white line in the
middle (that separates cyclists and pedestrians) could have
been done with highway=cycleway+footway=lane or something
similar. That is analogous to how we treat e.g. a tertiary
road with cycle lanes.
etc. etc. etc.
The path crowd however wanted "one solution for everything"
and can't accept that people didn't want to redo all existing
tagging. Especially not in places where it simply works.
The result is that some people use path as it is designed,
some people don't use path at all and other people use path
for what the translated word path means in their language
(often some kind of unpaved footway).
--
m.v.g.,
Cartinus
Nop
2009-11-30 07:36:25 UTC
Permalink
Hi!
Post by Steve Bennett
Before you go, do you think there is potential at least to have
consistency within each country?
It would be possible to solve the problem for each country.

It would also be possible to solve the problem generically for the whole
planet.

The real problem is that many people claim that there is no problem or
that they have already solved it and everybody should just do as they do.

Several of the approaches would work on their own if they were completed
to cover all use cases - but not with other interpretations using the
same tags in different ways thrown in between.

bye
Nop
Roy Wallace
2009-11-30 08:24:13 UTC
Permalink
Post by Nop
It would also be possible to solve the problem generically for the whole
planet.
The real problem is that many people claim that there is no problem or
that they have already solved it and everybody should just do as they do.
+1
Post by Nop
Several of the approaches would work on their own if they were completed
to cover all use cases - but not with other interpretations using the
same tags in different ways thrown in between.
+1. I wonder how to proceed...
Cartinus
2009-11-30 00:41:14 UTC
Permalink
Post by Nop
Richards view works only in the UK and fails
terribly in Germany and other countries.
Richards view works in a lot more countries than the UK. You can see it even
works in Germany by just looking at how Germany is currently mapped. Fuzzy
logic is flexible and extensible, that's why it works.

Where it doesn't work is in the minds of people who want one rigid solution
that solves everything in total detail. A solution that preferably looks like
some programming logic.
--
m.v.g.,
Cartinus
Nop
2009-11-30 07:29:22 UTC
Permalink
Hi!
Post by Cartinus
Post by Nop
Richards view works only in the UK and fails
terribly in Germany and other countries.
Richards view works in a lot more countries than the UK. You can see it even
works in Germany by just looking at how Germany is currently mapped. Fuzzy
logic is flexible and extensible, that's why it works.
Let me apply your logic to a different use case. Just imagine that in my
country there is a law that generally allows bicycles to use a one-way
road in both directions.

So I would define one-way as "mainly or exclusively intended for use in
one direction, bicycles may use both" and I claim that this is sufficient.

If you have a more rigid law where one-way is strictly for all vehicles,
it does not matter, fuzzy logic is good. Right?

I don't think so.

But again, it is a waste of time to discuss whether there is a problem
at all when we have chaotic and contradictory tagging for very basic use
cases. That is a problem.


bye
Nop
Cartinus
2009-11-30 07:45:10 UTC
Permalink
Post by Nop
Let me apply your logic to a different use case. Just imagine that in my
country there is a law that generally allows bicycles to use a one-way
road in both directions.
So I would define one-way as "mainly or exclusively intended for use in
one direction, bicycles may use both" and I claim that this is sufficient.
If you have a more rigid law where one-way is strictly for all vehicles,
? it does not matter, fuzzy logic is good. Right?
Yes, because there are two solutions to that "problem".

1) Add an extra tag in that single country that differs from the rest of the
world. But don't bother all the other mappers.

2) Any sufficiently sophisticated router will pre-process the data and it can
do something with different national defaults.
--
m.v.g.,
Cartinus


P.S. Gosmore ignores oneway for bicycle routing, but not for car routing. One
wonders why?
Steve Bennett
2009-11-30 08:02:15 UTC
Permalink
Post by Cartinus
Yes, because there are two solutions to that "problem".
1) Add an extra tag in that single country that differs from the rest of the
world. But don't bother all the other mappers.
IMHO "Don't piss off the whole world, just piss off one country" is a bad
solution, if there is no need to piss off anyone at all.
Post by Cartinus
2) Any sufficiently sophisticated router will pre-process the data and it can
do something with different national defaults.
Yes, but I would like us to define what the different national defaults are,
so that everyone can work off the same playbook.

For example, in Noppia, bikes can do the wrong one down one way streets. One
way streets are just tagged oneway, nothing special.
In Stevia, they can't.

We define use cases:
UC1) Oneway street with bikes allowed in wrong direction
UC2) Oneway street with bikes not allowed in wrong direction

We have a 2x2 matrix:

UC1
UC2
----------------------------------------------------------------
Noppia: oneway=true | oneway=true;bicycle=oneway
Stevia: oneway=true;bicycle=twoway | oneway=true

That's the table that needs to go in the wiki so that everyone understands
how to code the same thing in different countries.

Meanwhile the area for Noppia could be tagged
"bicycle_rule:wrong_way_in_oneway_permitted" or whatever.

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/3d85741f/attachment.html>
Roy Wallace
2009-11-30 08:32:17 UTC
Permalink
Post by Steve Bennett
IMHO "Don't piss off the whole world, just piss off one country" is a bad
solution, if there is no need to piss off anyone at all.
+1
Post by Steve Bennett
Yes, but I would like us to define what the different national defaults are,
so that everyone can work off the same playbook.
I'm not a fan of this solution, because usually I don't think it's
necessary - not in this case, anyway (read on...).
Post by Steve Bennett
For example, in Noppia, bikes can do the wrong one down one way streets. One
way streets are just tagged oneway, nothing special.
In Stevia, they can't.
UC1) Oneway street with bikes allowed in wrong direction
UC2) Oneway street with bikes not allowed in wrong direction
?????????????? UC1
UC2
----------------------------------------------------------------
Noppia: oneway=true????????????????? | oneway=true;bicycle=oneway
Stevia: oneway=true;bicycle=twoway?? | oneway=true
That's the table that needs to go in the wiki so that everyone understands
how to code the same thing in different countries.
Meanwhile the area for Noppia could be tagged
"bicycle_rule:wrong_way_in_oneway_permitted" or whatever.
I see your point, but WOW, that seems like a lot of extra STUFF to
maintain - and we don't have a good track record with maintenance (see
the wiki... :P). You don't need it. Use this, which is exactly as
*already documented in the wiki*:

UC1: oneway=yes; cycleway=opposite
UC2: oneway=yes

(see http://wiki.openstreetmap.org/wiki/Tag:oneway and
http://wiki.openstreetmap.org/wiki/Key:cycleway).
Steve Bennett
2009-11-30 10:06:20 UTC
Permalink
Post by Roy Wallace
Post by Steve Bennett
IMHO "Don't piss off the whole world, just piss off one country" is a bad
solution, if there is no need to piss off anyone at all.
+1
I see your point, but WOW, that seems like a lot of extra STUFF to
maintain - and we don't have a good track record with maintenance (see
the wiki... :P). You don't need it. Use this, which is exactly as
UC1: oneway=yes; cycleway=opposite
UC2: oneway=yes
You just pissed off Noppia. You just told them that every single oneway
street has to be explicitly marked "cycleway=opposite". The citizens of
Noppia resent this, and most of them refuse to put it in. "After all", they
reason, "everyone knows that you can ride the wrong way up any oneway
street".

And you reply...?

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/f0fee96a/attachment.html>
Roy Wallace
2009-11-30 10:54:03 UTC
Permalink
Post by Steve Bennett
Post by Roy Wallace
Post by Steve Bennett
IMHO "Don't piss off the whole world, just piss off one country" is a bad
solution, if there is no need to piss off anyone at all.
+1
I see your point, but WOW, that seems like a lot of extra STUFF to
maintain - and we don't have a good track record with maintenance (see
the wiki... :P). You don't need it. Use this, which is exactly as
UC1: oneway=yes; cycleway=opposite
UC2: oneway=yes
You just pissed off Noppia. You just told them that every single oneway
street has to be explicitly marked "cycleway=opposite". The citizens of
Noppia resent this, and most of them refuse to put it in. "After all", they
reason, "everyone knows that you can ride the wrong way up any oneway
street".
And you reply...?
Hmm good question...

Several thoughts:

1) I told them that *the wiki recommends* that they do need to use
cycleway=opposite where appropriate.

1a) This is different to *me* telling them what to do - the wiki
carries more weight as it is the outcome of discussion (see the
discussion page, for example). It's also where newbies go to learn how
to map, and where others (me, at least) go for reference. Using a
common set of guidelines like this is key to maintaining consistency.
Also, importantly, if the Noppians think something is suboptimal in
the wiki, and want to re-open the discussion, propose something else,
etc., there are mechanisms available for that.

1b) Is it really so hard to add cycleway=opposite where applicable?
Really? Maybe I'm missing something (but then again, I'm one of those
strange people who have no problem adding source=* tags to everything
I change). I am always a little perplexed at some people's aversion to
extra tags - we have autocomplete, presets, DB compression, ... I
don't think it is ever worth compromising consistency to save
keystrokes.

2) They may think "everyone" knows the rules in Noppia, but this is
unlikely to be true. e.g. what if I visit Noppia on holiday or
business? What if my routing software uses the defaults for oneway=*
as described in the wiki?

3) You say the citizens "refuse" to follow the wiki's recommendations.
If they do realise that this is a problem, I cannot imagine that they
would refuse to change their practices - after all, usually OSM
contributors do want to contribute to a consistent i.e. useful OSM
database. If they can't see that ignoring the wiki can be dangerous,
then I would probably leave the room in frustration.

But Steve, the point is that surely the Noppians also want to come up
with a solution that gives us the best possible OSM database. Right? I
would ask them: what do they think is the best way to achieve that?
Steve Bennett
2009-11-30 11:08:03 UTC
Permalink
Post by Roy Wallace
1) I told them that *the wiki recommends* that they do need to use
cycleway=opposite where appropriate.
1a) This is different to *me* telling them what to do - the wiki
carries more weight as it is the outcome of discussion (see the
discussion page, for example). It's also where newbies go to learn how
Ok, to recap, I said
Post by Roy Wallace
IMHO "Don't piss off the whole world, just piss off one country" is a bad
solution, if there is no need to piss off anyone at all.
+1
(It's ok if you don't agree with the position, but you're having it both
ways - saying you don't want to piss off Noppia, but then complaining when
the Noppians get pissed off.

to map, and where others (me, at least) go for reference. Using a
Post by Roy Wallace
common set of guidelines like this is key to maintaining consistency.
Across all countries? Why?
Post by Roy Wallace
Also, importantly, if the Noppians think something is suboptimal in
the wiki, and want to re-open the discussion, propose something else,
etc., there are mechanisms available for that.
But they're only one country. As long as there is an assumption that
everyone has to follow the same rules, then there are going to be losers.
Post by Roy Wallace
1b) Is it really so hard to add cycleway=opposite where applicable?
Yep. Like I said, I refuse to add "direction=clockwise" to mini_roundabouts.
One extra tag is a lot of extra effort, when the total number of tags you're
adding is usually 2-3.
Post by Roy Wallace
Really? Maybe I'm missing something (but then again, I'm one of those
strange people who have no problem adding source=* tags to everything
I change). I am always a little perplexed at some people's aversion to
extra tags - we have autocomplete, presets, DB compression, ... I
don't think it is ever worth compromising consistency to save
keystrokes.
What if you had to type "bicycle=yes" on every single road? It would suck.
How about "car=yes"? How about
"bicycle=yes;car=yes;bus=yes;surface=paved;smoothness=5;colour=black;lines=white;parking=parallel;lanes=2;"
on every road?

2) They may think "everyone" knows the rules in Noppia, but this is
Post by Roy Wallace
unlikely to be true. e.g. what if I visit Noppia on holiday or
business?
What if my routing software uses the defaults for oneway=*
as described in the wiki?
Then your routing software needs to be Noppia-compatible. And since,
theoretically, we have published an RFC explaining all the international
variations in an XML file, what the rules are, that's easy.
Post by Roy Wallace
3) You say the citizens "refuse" to follow the wiki's recommendations.
If they do realise that this is a problem, I cannot imagine that they
They don't have a problem. Their maps render fine, and they know exactly how
oneway streets.
Post by Roy Wallace
would refuse to change their practices - after all, usually OSM
contributors do want to contribute to a consistent i.e. useful OSM
Locally consistent.
Post by Roy Wallace
But Steve, the point is that surely the Noppians also want to come up
with a solution that gives us the best possible OSM database. Right? I
would ask them: what do they think is the best way to achieve that?
That's one goal. I suspect most people's goals are more pragmatic and
localised. Do I really care what the Bulgarian OSM data looks like? Not
unless I'm going there. Do I care what the Melbourne data looks like? Yes.
Do I care what the Melbourne bike path data looks like? Yes, a lot.

Am I out of line here? Of course I want to see a globally consistent, useful
database. But ultimately, I want to see the most number of users happy with
their local data. And if that means tags mean something slightly different
in Cambodia than they do in Ireland, then...what was the problem again?

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/ef5d3d15/attachment.html>
Roy Wallace
2009-11-30 11:28:15 UTC
Permalink
Post by Steve Bennett
Am I out of line here? Of course I want to see a globally consistent, useful
database. But ultimately, I want to see the most number of users happy with
their local data. And if that means tags mean something slightly different
in Cambodia than they do in Ireland, then...what was the problem again?
Ok, let me summarise my position, before this thread derails.

I think we should aim for a globally consistent database, because
1) I travel a fair bit (I've never been to Bulgaria, but maybe someday soon)
2) I do NOT want to be limited to "Noppia-compatible" routing software
if I visit Noppia (etc.)
3) I think it's not that hard to be globally consistent - it just
comes at the cost of verbosity (which is cheap)

Adding tags that help clarify what I mean does not piss me off. I am
quite happy to add direction=clockwise to roundabouts if necessary.
Ultimately, why not aim to have direction=* applied to ALL
roundabouts? I know you have a different position, which is fine. I'm
surprised that you feel "one extra tag is a lot of extra effort" -
have you tried various editor presets, auto-complete, selecting
multiple entities before applying a tag, etc.?

For me, your example of a road tagged with:

"bicycle=yes;car=yes;bus=yes;surface=paved;smoothness=5;colour=black;lines=white;parking=parallel;lanes=2"

just looks like a very well-mapped road. "Good job", I would say to
the mapper, as they were obviously very thorough. Seriously.
Steve Bennett
2009-11-30 11:51:21 UTC
Permalink
Post by Roy Wallace
I think we should aim for a globally consistent database, because
1) I travel a fair bit (I've never been to Bulgaria, but maybe someday soon)
2) I do NOT want to be limited to "Noppia-compatible" routing software
if I visit Noppia (etc.)
Consider it "internationally aware" software. Routing software that is aware
of the local laws of each country seems obvious. We'll limit the variations
as much as possible, of course.
Post by Roy Wallace
3) I think it's not that hard to be globally consistent - it just
comes at the cost of verbosity (which is cheap)
I think verbosity is expensive. My experience is with Wikipedia, where
everyone always thinks the labour is free. It may be free, but it's finite.
And the more you get people to waste their time doing tedious busywork, the
less time they spend doing useful things.
Post by Roy Wallace
Adding tags that help clarify what I mean does not piss me off. I am
quite happy to add direction=clockwise to roundabouts if necessary.
Ultimately, why not aim to have direction=* applied to ALL
roundabouts?
Sure, by all means, have that tag applied. But forcing someone to manually
add it when the roundabout in question is in a left-drive country is
insulting. Maybe the client could add it automatically. I don't know.
Post by Roy Wallace
I know you have a different position, which is fine. I'm
surprised that you feel "one extra tag is a lot of extra effort" -
have you tried various editor presets, auto-complete, selecting
multiple entities before applying a tag, etc.?
Auto-complete, yes, and I still think plus-s-o-enter-n-enter is too many
keystrokes to add "source=nearmap". (It's even worse in josm:
alt+b-s-o-tab-n-enter).

Maybe I need to use josm to do something like search for everything I've
touched that has no source and bulk-update. But that could make mistakes.

Will investigate "editor presets".

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/564ec1c2/attachment.html>
Roy Wallace
2009-11-30 12:38:04 UTC
Permalink
Post by Roy Wallace
I think we should aim for a globally consistent database, because
1) I travel a fair bit (I've never been to Bulgaria, but maybe someday soon)
2) I do NOT want to be limited to "Noppia-compatible" routing software
if I visit Noppia (etc.)
Consider it "internationally aware" software. Routing software that is aware of the local laws of each country seems obvious.
Um...what??? That will not write itself. Do you expect us to
successfully digitize and maintain a database of all laws of all
countries? In a wiki, even? That's ambitious! I'd prefer to stick to
mapping what's on the ground.
Post by Roy Wallace
3) I think it's not that hard to be globally consistent - it just
comes at the cost of verbosity (which is cheap)
I think verbosity is expensive. My experience is with Wikipedia, where
everyone always thinks the labour is free. It may be free, but it's finite.
And the more you get people to waste their time doing tedious busywork, the
less time they spend doing useful things.
I agree that tedious busywork is not good. But we have computers -
surely we're able to use presets etc. so that more
verbosity/explicitness requires negligible amounts of additional
labour. Let's get the tagging schemes right first. Seriously, it's not
going to be a big deal to e.g. add foot=yes/no to cycleways. Look at
the big picture - we're making a map of the entire world. We're trying
to find the best and easiest way to do it. Remember that additional
labour adding foot=yes/no can *avoid* future labour spent sorting out
messes like this one. And it can give us a better quality result.
Post by Roy Wallace
Adding tags that help clarify what I mean does not piss me off. I am
quite happy to add direction=clockwise to roundabouts if necessary.
Ultimately, why not aim to have direction=* applied to ALL
roundabouts?
Sure, by all means, have that tag applied. But forcing someone to manually
add it when the roundabout in question is in a left-drive country is
insulting. Maybe the client could add it automatically. I don't know.
Well, I don't find it insulting. And yes, the "client" (editor) could
certainly add it automatically. Remember that we are also not limited
to current versions of current editors - editors can be improved.

Let's get the tagging right first - editor improvements will follow. I
think we shouldn't "tag for the editor" (if you know what I mean) :P
Steve Bennett
2009-11-30 12:47:22 UTC
Permalink
Post by Roy Wallace
Um...what??? That will not write itself. Do you expect us to
successfully digitize and maintain a database of all laws of all
countries?
What do you think? Work with me, here.
Post by Roy Wallace
In a wiki, even? That's ambitious! I'd prefer to stick to
mapping what's on the ground.
I agree that tedious busywork is not good. But we have computers -
surely we're able to use presets etc. so that more
verbosity/explicitness requires negligible amounts of additional
labour.
Yes...macros and scripts are always a first step in improving usability.
Smarter data structures and algorithms, and better analysis of needs and
solutions is the next.
Post by Roy Wallace
Let's get the tagging schemes right first. Seriously, it's not
going to be a big deal to e.g. add foot=yes/no to cycleways.
You: It's easy to add "foot=yes".
Me: It's hard to get everyone to consistently add "foot=yes".

Just so we're clear on that. Can we move on?
Post by Roy Wallace
Let's get the tagging right first - editor improvements will follow.
If by "get the tagging right" you mean "analyse the problem, work out what
people are doing, and come up with the most efficient set of tags for people
to use", then yes. But I don't think you mean that.

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/63e2b8b5/attachment.html>
Roy Wallace
2009-11-30 21:25:36 UTC
Permalink
Post by Steve Bennett
Post by Roy Wallace
Um...what??? That will not write itself. Do you expect us to
successfully digitize and maintain a database of all laws of all
countries?
What do you think? Work with me, here.
I think that would be a nightmare, and would not work. If anything, it
would introduce MORE inconsistency due to 1) difficulty maintaining
the lawbook and 2) a more complicated set of guidelines and more
complicated wiki, making it even LESS likely that people will follow
it consistently.

As I've said, I'd prefer to stick to *mapping what's on the ground*,
*according to the guidelines in the wiki*. This is the only way to get
global consistency, which I think is important for the reasons I've
already described.
Post by Steve Bennett
Post by Roy Wallace
Let's get the tagging schemes right first. Seriously, it's not
going to be a big deal to e.g. add foot=yes/no to cycleways.
You: It's easy to add "foot=yes".
Me: It's hard to get everyone to consistently add "foot=yes".
Just so we're clear on that. Can we move on?
Me: It *will be* easy to get everyone to consistently add "foot=yes" when:
1) I can convince you guys that this approach is the best way to get
global consistency, and that that's important;
2) people realise that editors can be used to avoid additional
keystrokes and so there is actually no cost in adding foot=yes;
3) this mess is sorted out, and the guidelines for
path/footway/cycleway are consolidated and improved (made clear)

Re: 3), I often hear people say "it's such a mess, I gave up asking on
the email list and now I just use cycleway when [ insert custom
definition ]".
Post by Steve Bennett
Post by Roy Wallace
Let's get the tagging right first - editor improvements will follow.
If by "get the tagging right" you mean "analyse the problem, work out what
people are doing, and come up with the most efficient set of tags for people
to use", then yes. But I don't think you mean that.
I do mean that! Assuming that, by "most efficient", you mean "most
likely to result in a complete and consistent map of the Earth". And
before you say "but that's not necessarily efficient", part of being
"likely to result in a good outcome" is that mappers remain motivated
to contribute - so this does take into account that the tags have to
be satisfying to use.
Cartinus
2009-11-30 21:55:31 UTC
Permalink
Post by Roy Wallace
1) I can convince you guys that this approach is the best way to get
global consistency, and that that's important;
2) people realise that editors can be used to avoid additional
keystrokes and so there is actually no cost in adding foot=yes;
I've been told that when OSM started (I wasn't involved then) that every
motorway had to be tagged horse=no+foot=no+bicycle=no.

There is a reason they stopped doing that.
--
m.v.g.,
Cartinus
Roy Wallace
2009-11-30 22:43:48 UTC
Permalink
Post by Cartinus
Post by Roy Wallace
1) I can convince you guys that this approach is the best way to get
global consistency, and that that's important;
2) people realise that editors can be used to avoid additional
keystrokes and so there is actually no cost in adding foot=yes;
I've been told that when OSM started (I wasn't involved then) that every
motorway had to be tagged horse=no+foot=no+bicycle=no.
There is a reason they stopped doing that.
The reason is that that's *globally* redundant.
Liz
2009-11-30 22:57:56 UTC
Permalink
Post by Roy Wallace
Post by Cartinus
Post by Roy Wallace
1) I can convince you guys that this approach is the best way to get
global consistency, and that that's important;
2) people realise that editors can be used to avoid additional
keystrokes and so there is actually no cost in adding foot=yes;
I've been told that when OSM started (I wasn't involved then) that every
motorway had to be tagged horse=no+foot=no+bicycle=no.
There is a reason they stopped doing that.
The reason is that that's *globally* redundant.
not exactly correct.
We do have highway marked motorway in Au where bicycles are allowed.
Roy Wallace
2009-11-30 23:04:55 UTC
Permalink
Post by Liz
Post by Roy Wallace
Post by Cartinus
Post by Roy Wallace
1) I can convince you guys that this approach is the best way to get
global consistency, and that that's important;
2) people realise that editors can be used to avoid additional
keystrokes and so there is actually no cost in adding foot=yes;
I've been told that when OSM started (I wasn't involved then) that every
motorway had to be tagged horse=no+foot=no+bicycle=no.
There is a reason they stopped doing that.
The reason is that that's *globally* redundant.
not exactly correct.
We do have highway marked motorway in Au where bicycles are allowed.
Ok, rephrased: the reason they stopped is because it wasn't necessary.
Obviously, we have a problem here. I'm suggesting some solutions.
Elizabeth Dodd
2009-11-30 23:13:51 UTC
Permalink
Post by Roy Wallace
Post by Liz
Post by Roy Wallace
Post by Cartinus
Post by Roy Wallace
1) I can convince you guys that this approach is the best way to get
global consistency, and that that's important;
2) people realise that editors can be used to avoid additional
keystrokes and so there is actually no cost in adding foot=yes;
I've been told that when OSM started (I wasn't involved then) that
every motorway had to be tagged horse=no+foot=no+bicycle=no.
There is a reason they stopped doing that.
The reason is that that's *globally* redundant.
not exactly correct.
We do have highway marked motorway in Au where bicycles are allowed.
Ok, rephrased: the reason they stopped is because it wasn't necessary.
Obviously, we have a problem here. I'm suggesting some solutions.
I'm not sure that those roads (Hume Highway) should be marked as motorway, but
got no comment on the talk-au list when i asked for comments.
Steve Bennett
2009-11-30 23:42:43 UTC
Permalink
Various comments:

I'm not sure that those roads (Hume Highway) should be marked as motorway,
Post by Elizabeth Dodd
but
got no comment on the talk-au list when i asked for comments.
The Hume *Freeway* is definitely a motorway. There are places between
Melbourne and Sydney where it's just a highway, but it's dual carriage
almost the whole way through Victoria. The Western Freeway also allows
bikes. (And for anyone who hasn't tried it, riding on a motorway has some
serious slipstreaming benefits!)
Post by Elizabeth Dodd
This is one area where national defaults won't help - you'd have to get
down to local rules.

I've referred to "Jurisdictions" a few times for this reason. I imagine US
states are possibly even more individual. Would we go as far as
councils/municipalities? Probably not. (Although, as I mentioned somewhere
earlier, the City of Melbourne prohibits bike riding in parks, while other
councils don't. My preferred solution would be tag paths in the former as
footway, and in the latter as cycleway).
Post by Elizabeth Dodd
The point I was making was that it should *not* be necessary to
*require* "a database of all laws of all countries"
This statement is unnecessarily scary. "Database". "All laws". "All
countries". Like I said, think about it. Of course not. The variations are
minor, and represent a tiny proportion of the traffic laws of each country.
And we could certainly have defaults that most countries match.

Remember, the goal here is to have a situation where the most intuitive and
convenient way to tag stuff where people live is as consistent as possible
with the rest of the world. A cultural/legal matrix is the right way to do
that.

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091201/e8dd55ed/attachment.html>
Liz
2009-12-01 00:04:17 UTC
Permalink
Post by Elizabeth Dodd
I'm not sure that those roads (Hume Highway) should be marked as motorway,
Post by Elizabeth Dodd
but
got no comment on the talk-au list when i asked for comments.
The Hume Freeway is definitely a motorway. There are places between
Melbourne and Sydney where it's just a highway, but it's dual carriage
almost the whole way through Victoria. The Western Freeway also allows
bikes. (And for anyone who hasn't tried it, riding on a motorway has some
serious slipstreaming benefits!)
i agree with you
there are many km of Hume Highway in NSW which are marked as motorway, when
they are technically not so, they are a high quality dual carriageway. But
there are no flyovers, traffic can cross the lanes to enter and leave the
road.
Martin Koppenhoefer
2009-12-01 00:09:36 UTC
Permalink
2009/12/1 Steve Bennett <stevagewp at gmail.com>
Post by Steve Bennett
I've referred to "Jurisdictions" a few times for this reason. I imagine US
states are possibly even more individual. Would we go as far as
councils/municipalities? Probably not. (Although, as I mentioned somewhere
earlier, the City of Melbourne prohibits bike riding in parks, while other
councils don't. My preferred solution would be tag paths in the former as
footway, and in the latter as cycleway).
no, I wouldn't tag all ways in a park as cycleways, just because you can
also ride a bike. I would tag both as footway (or service if big enough for
cars to drive, with access-tags) and put bicycle=yes/no aside, as long as
they are not dedicated cycleways.

cheers,
Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091201/3acb52a3/attachment.html>
Anthony
2009-12-01 01:03:54 UTC
Permalink
Post by Roy Wallace
Post by Steve Bennett
Post by Roy Wallace
Um...what??? That will not write itself. Do you expect us to
successfully digitize and maintain a database of all laws of all
countries?
What do you think? Work with me, here.
I think that would be a nightmare, and would not work. If anything, it
would introduce MORE inconsistency due to 1) difficulty maintaining
the lawbook and 2) a more complicated set of guidelines and more
complicated wiki, making it even LESS likely that people will follow
it consistently.
As I've said, I'd prefer to stick to *mapping what's on the ground*,
*according to the guidelines in the wiki*. This is the only way to get
global consistency, which I think is important for the reasons I've
already described.
If the law of one jurisdiction says bicycles are allowed on all roads
except freeways, and the law of another jurisdiction says bicycles are
allowed on all roadways with speed limits less than 45, and there
aren't any signs "on the ground" making people aware of this, and we
don't want to maintain a database of laws, what are we to do?

I like the idea of mapping only "what's on the ground". But it can be
taken to far. Does anyone honestly suggest that we shouldn't tag a
road's name using any knowledge other than what's literally "on the
ground" - on a street sign? Should we really do away with pretty much
all boundary=* tags altogether (or replace them with nodes at the
points where there are signs)? "Map what's on the ground" is a good
guideline, but there have to be exceptions.

Tobias Knerr
2009-11-30 22:16:37 UTC
Permalink
Post by Roy Wallace
Routing software that is aware of the local laws of each country seems obvious.
Um...what??? That will not write itself. Do you expect us to
successfully digitize and maintain a database of all laws of all
countries? In a wiki, even? That's ambitious! I'd prefer to stick to
mapping what's on the ground.
If we map what's on the ground, then we create a map database containing
"here is an oneway sign, over there is a cycleway sign". That's
nice, but if I want to do routing with this, I need information such as
"can I use way w in direction d with vehicle v?" - and in order to know
this, I need another database that tells me what a sign means in that
part of the world (for example: are pedestrians allowed to walk on ways
with a cycleway sign?).

If we don't want a traffic law database, then we need to tag the
required information directly. But then mappers don't just map physical
reality. They interpret the signs (and other information) using their -
hopefully correct - knowledge of the laws.

Both can work, but /someone/ has to do the transfer from reality to road
network attributes - either software (using a traffic laws DB) or humans
(mapping more than just what's on the ground).

Tobias Knerr
Roy Wallace
2009-11-30 23:03:15 UTC
Permalink
Post by Tobias Knerr
Post by Roy Wallace
Routing software that is aware of the local laws of each country seems obvious.
Um...what??? That will not write itself. Do you expect us to
successfully digitize and maintain a database of all laws of all
countries? In a wiki, even? That's ambitious! I'd prefer to stick to
mapping what's on the ground.
If we map what's on the ground, then we create a map database containing
"here is an oneway sign, over there is a cycleway sign". That's
nice, but if I want to do routing with this, I need information such as
"can I use way w in direction d with vehicle v?" - and in order to know
this, I need another database that tells me what a sign means in that
part of the world (for example: are pedestrians allowed to walk on ways
with a cycleway sign?).
If we don't want a traffic law database, then we need to tag the
required information directly. But then mappers don't just map physical
reality. They interpret the signs (and other information) using their -
hopefully correct - knowledge of the laws.
Both can work, but /someone/ has to do the transfer from reality to road
network attributes - either software (using a traffic laws DB) or humans
(mapping more than just what's on the ground).
Good points. You did find a flaw in my argument - that I was sort of
advocating exhaustive tagging as well as only mapping what's on the
ground. Funnily enough, I actually find both of these extremes
acceptable. But that's not the point...

The point I was making was that it should *not* be necessary to
*require* "a database of all laws of all countries" to know what
highway=cycleway means. There should be one definition that is
consistent for the whole world. For example, "this path is marked with
a sign with a bicycle symbol on it". If people also want to put in
exhaustive information inferred from a law book, I'd prefer they go
ahead and use "foot=no + source:foot=lawbook". If people prefer to
leave out the inferred information, and instead write routers with
country-specific defaults, that's cool, too.

But highway=cycleway tags in the OSM database should all mean the same thing.
Tobias Knerr
2009-11-30 23:34:24 UTC
Permalink
Post by Roy Wallace
The point I was making was that it should *not* be necessary to
*require* "a database of all laws of all countries" to know what
highway=cycleway means. There should be one definition that is
consistent for the whole world. For example, "this path is marked with
a sign with a bicycle symbol on it". If people also want to put in
exhaustive information inferred from a law book, I'd prefer they go
ahead and use "foot=no + source:foot=lawbook". If people prefer to
leave out the inferred information, and instead write routers with
country-specific defaults, that's cool, too.
But highway=cycleway tags in the OSM database should all mean the same thing.
Do you only suggest that there should be exactly one meaning per tag, or
would you also want the same tags to be used all over the world?

It makes a difference for possible approaches like using
highway=Fahrradweg (or DE:cycleway or any other value that isn't exactly
"cycleway") for German cycleways, as that would still be "one meaning
per tag".

Tobias Knerr
Roy Wallace
2009-11-30 23:51:48 UTC
Permalink
Post by Tobias Knerr
Post by Roy Wallace
The point I was making was that it should *not* be necessary to
*require* "a database of all laws of all countries" to know what
highway=cycleway means. There should be one definition that is
consistent for the whole world. For example, "this path is marked with
a sign with a bicycle symbol on it". If people also want to put in
exhaustive information inferred from a law book, I'd prefer they go
ahead and use "foot=no + source:foot=lawbook". If people prefer to
leave out the inferred information, and instead write routers with
country-specific defaults, that's cool, too.
But highway=cycleway tags in the OSM database should all mean the same thing.
Do you only suggest that there should be exactly one meaning per tag, or
would you also want the same tags to be used all over the world?
It makes a difference for possible approaches like using
highway=Fahrradweg (or DE:cycleway or any other value that isn't exactly
"cycleway") for German cycleways, as that would still be "one meaning
per tag".
One meaning per tag is essential. If a German cycleway is *different*
in some important way to a UK (or whatever) cycleway, it should
ultimately be tagged *differently*. I find this obvious.

The details of how they should be tagged are of secondary importance.
To answer your question specifically, I don't have so much of a
problem with highway=fahrradweg, though I doubt that's the best way to
do it. If highway=fahrradweg at least has the same meaning in all
usages, it would be an improvement!
Martin Koppenhoefer
2009-12-01 00:14:23 UTC
Permalink
2009/12/1 Roy Wallace <waldo000000 at gmail.com>
Post by Roy Wallace
One meaning per tag is essential.
it depends what this meaning is. If you intend by meaning: cycleway is a way
with a bicycle-sign: fine, if you intent that all access rights should be
implicitly and globally given: no.
Post by Roy Wallace
If a German cycleway is *different*
in some important way to a UK (or whatever) cycleway, it should
ultimately be tagged *differently*. I find this obvious.
what's the difference? Minimum width differs 5 cm? Kind of sign? Forbidden
to pedestrians? Obligatory for bicycles? Forbidden to 25ccm? Blue lines
instead of white ones? If every smallest difference will cause another
top-tag, we'll get thousands of them.

cheers,
Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091201/0486ca11/attachment.html>
Martin Koppenhoefer
2009-12-01 00:07:37 UTC
Permalink
2009/12/1 Tobias Knerr <osm at tobias-knerr.de>
Post by Tobias Knerr
Post by Roy Wallace
The point I was making was that it should *not* be necessary to
*require* "a database of all laws of all countries" to know what
highway=cycleway means.
+1. even if for implicit regulations this would be needed, at least the
consensus could be (like you suggested) that cycleway is a marked cycleway.
Whether I'm allowed to use it with a 25ccm motorcycle (mofa) or not, could
be optionally tagged or considered as implicitly given by local law.
Post by Tobias Knerr
Do you only suggest that there should be exactly one meaning per tag, or
would you also want the same tags to be used all over the world?
what do you mean by this? The meaning is always different. Do we need
different tags for roads in the UK because the drive on the left? IMHO no.

cheers,
Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091201/287010b3/attachment.html>
Anthony
2009-12-01 00:21:27 UTC
Permalink
Post by Roy Wallace
That will not write itself. Do you expect us to
successfully digitize and maintain a database of all laws of all
countries? In a wiki, even? That's ambitious! I'd prefer to stick to
mapping what's on the ground.
You can map what's on the ground, then. But in order to make a decent
routing application, someone is going to have to maintain a database
of certain laws in any states they wish for their routing application
to work. There's really no question about that. The only question is
whether you want to store that information in every single element of
the database, or store it once for each jurisdiction.

I vote for once for each jurisdiction. But I vote strongly against
doing so using a wiki.
Frederik Ramm
2009-12-01 00:37:05 UTC
Permalink
Hi,
Post by Anthony
But in order to make a decent
routing application, someone is going to have to maintain a database
of certain laws in any states they wish for their routing application
to work.
It is certainly good to know what is allowed.

But a good routing application should also consider (and I think this
was recently mentioned by someone else) the physically possible which
might be more or less than what's legal (there are over-cautios and
litigious jurisdictions where the allowed is often a subset of the
possible, and there are laid-back jurisdictions where there's no speed
limit but if you go too fast you die and rot).

And then both axes are not really "boolean". Between the physically
possible and the physically impossible may lie an area that requires
more skill, better vehicles or simply means a higher risk of accidents.
Between the allowed and the forbidden lie several steps of badness - how
likely is it that I am found out, and what fine or punishment am I in
for if I am found out?

A good routing application will lay this wealth of information out
before you, so that you can decide whether you'd rather risk injury,
penalty and being re-born as a rat, but save time and fuel, or whether
you prefer to pay a little more for safety.

Existing commercial routing applications take the easy way out by
excluding anything that is not legal. I hope we won't!

Bye
Frederik
--
Frederik Ramm ## eMail frederik at remote.org ## N49?00'09" E008?23'33"
Martin Koppenhoefer
2009-11-30 11:25:57 UTC
Permalink
2009/11/29 Roy Wallace <waldo000000 at gmail.com>
Post by Roy Wallace
Post by Anthony
When is there a path and when is there not a path? I walk through an
area of grass every time I go to the park near my house. Isn't that a
"path" which is part of "reality"?
An area of grass is - to me - not a path. A path, IMHO, is something
that exists independently of people walking or not walking on it (i.e.
usually you can *see* that it resembles a path).
-1, a path is either planned and constructed (the ones you are refering to)
or it "creates itself" by frequent use (e.g. shortcuts on grass). IMHO the
latter are even more valueable to the project because they are usable but
you don't find them in other maps.

cheers,
Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/1931972f/attachment.html>
Roy Wallace
2009-11-30 11:39:17 UTC
Permalink
On Mon, Nov 30, 2009 at 9:25 PM, Martin Koppenhoefer
Post by Martin Koppenhoefer
Post by Roy Wallace
An area of grass is - to me - not a path. A path, IMHO, is something
that exists independently of people walking or not walking on it (i.e.
usually you can *see* that it resembles a path).
-1, a path is either planned and constructed (the ones you are refering to)
or it "creates itself" by frequent use (e.g. shortcuts on grass). IMHO the
latter are even more valueable to the project because they are usable but
you don't find them in other maps.
A shortcut through grass that you can see, sure! e.g.
http://s0.geograph.org.uk/photos/18/97/189701_92c9a5d5.jpg

But if you can't see it - sorry - you're not going to convince me that
there is a path.

If you can see some grass, sure, map that. But just being able to walk
on the grass does not turn the grass into a path. Otherwise, in any
area of grass there would actually be *infinite* overlapping,
criss-crossing "invisible-paths". :P
Peter Childs
2009-11-30 11:50:44 UTC
Permalink
Post by Roy Wallace
On Mon, Nov 30, 2009 at 9:25 PM, Martin Koppenhoefer
Post by Martin Koppenhoefer
Post by Roy Wallace
An area of grass is - to me - not a path. A path, IMHO, is something
that exists independently of people walking or not walking on it (i.e.
usually you can *see* that it resembles a path).
-1, a path is either planned and constructed (the ones you are refering to)
or it "creates itself" by frequent use (e.g. shortcuts on grass). IMHO the
latter are even more valueable to the project because they are usable but
you don't find them in other maps.
A shortcut through grass that you can see, sure! e.g.
http://s0.geograph.org.uk/photos/18/97/189701_92c9a5d5.jpg
But if you can't see it - sorry - you're not going to convince me that
there is a path.
If you can see some grass, sure, map that. But just being able to walk
on the grass does not turn the grass into a path. Otherwise, in any
area of grass there would actually be *infinite* overlapping,
criss-crossing "invisible-paths". :P
Perhaps what we need here is a tag that says you can walk anyway you
like within this area, Like a large town squares, playing field, etc I
know that places like Scotland there is a "Right to Roam" but for most
of us, we need to keep to paths but sometimes areas are less
strict.... Walking routing software could see this area and take the
shortest route across the area.

Peter
Martin Koppenhoefer
2009-11-30 12:00:01 UTC
Permalink
2009/11/30 Roy Wallace <waldo000000 at gmail.com>
Post by Roy Wallace
A shortcut through grass that you can see, sure! e.g.
http://s0.geograph.org.uk/photos/18/97/189701_92c9a5d5.jpg
But if you can't see it - sorry - you're not going to convince me that
there is a path.
+1, I completely agree with you. Only visible paths (where visibility
indicates frequent use, if it is not in use, there won't be a visible
trail). I guess I misunderstood you before.

cheers,
Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/3f145ec2/attachment.html>
Anthony
2009-11-30 16:08:20 UTC
Permalink
Post by Roy Wallace
On Mon, Nov 30, 2009 at 9:25 PM, Martin Koppenhoefer
Post by Martin Koppenhoefer
Post by Roy Wallace
An area of grass is - to me - not a path. A path, IMHO, is something
that exists independently of people walking or not walking on it (i.e.
usually you can *see* that it resembles a path).
-1, a path is either planned and constructed (the ones you are refering to)
or it "creates itself" by frequent use (e.g. shortcuts on grass). IMHO the
latter are even more valueable to the project because they are usable but
you don't find them in other maps.
A shortcut through grass that you can see, sure! e.g.
http://s0.geograph.org.uk/photos/18/97/189701_92c9a5d5.jpg
But if you can't see it - sorry - you're not going to convince me that
there is a path.
If you can see some grass, sure, map that. But just being able to walk
on the grass does not turn the grass into a path. Otherwise, in any
area of grass there would actually be *infinite* overlapping,
criss-crossing "invisible-paths". :P
What if I map the entire section of grass which is within the right of
way as a polygon with highway=path, area=yes? That's how we represent
infinite overlapping criss-crossing "invisible-paths", like a
pedestrian mall.

On the right is a road. On the left is a lake. In the middle, is a
path, made out of grass. It's probably not much wider than the road.
And only about half of it is within the right of way.
Steve Bennett
2009-11-30 16:20:38 UTC
Permalink
Post by Anthony
What if I map the entire section of grass which is within the right of
way as a polygon with highway=path, area=yes? That's how we represent
infinite overlapping criss-crossing "invisible-paths", like a
pedestrian mall.
I'm kind of hoping future routers will assume people can walk anywhere
within parks, if it saves time. For most of the parks I've been dealing
with, it would make far more sense to map the occasional barrier rather than
all the open space.

(I confess I've been dreaming of a super-router that plots a path, then
quickly walks you through it, checking that you're happy with its
decisions...)

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091201/845d0043/attachment.html>
Anthony
2009-11-30 16:22:41 UTC
Permalink
Post by Steve Bennett
Post by Anthony
What if I map the entire section of grass which is within the right of
way as a polygon with highway=path, area=yes? ?That's how we represent
infinite overlapping criss-crossing "invisible-paths", like a
pedestrian mall.
I'm kind of hoping future routers will assume people can walk anywhere
within parks, if it saves time. For most of the parks I've been dealing
with, it would make far more sense to map the occasional barrier rather than
all the open space.
Maybe, but this isn't in the park. This is on the way to the park.
The "way". Ha. No pun intended.
Lester Caine
2009-11-30 16:36:14 UTC
Permalink
Post by Anthony
Post by Roy Wallace
On Mon, Nov 30, 2009 at 9:25 PM, Martin Koppenhoefer
Post by Martin Koppenhoefer
Post by Roy Wallace
An area of grass is - to me - not a path. A path, IMHO, is something
that exists independently of people walking or not walking on it (i.e.
usually you can *see* that it resembles a path).
-1, a path is either planned and constructed (the ones you are refering to)
or it "creates itself" by frequent use (e.g. shortcuts on grass). IMHO the
latter are even more valueable to the project because they are usable but
you don't find them in other maps.
A shortcut through grass that you can see, sure! e.g.
http://s0.geograph.org.uk/photos/18/97/189701_92c9a5d5.jpg
But if you can't see it - sorry - you're not going to convince me that
there is a path.
If you can see some grass, sure, map that. But just being able to walk
on the grass does not turn the grass into a path. Otherwise, in any
area of grass there would actually be *infinite* overlapping,
criss-crossing "invisible-paths". :P
What if I map the entire section of grass which is within the right of
way as a polygon with highway=path, area=yes? That's how we represent
infinite overlapping criss-crossing "invisible-paths", like a
pedestrian mall.
On the right is a road. On the left is a lake. In the middle, is a
path, made out of grass. It's probably not much wider than the road.
And only about half of it is within the right of way.
I think that is probably covers some of the 'paths' that I need to describe.
They are really large areas of grass which can be walked across to get to other
points, rather than a 'tag' on the side of a near by road .... using the road to
do the 'pedestrian' routing is simply wrong but so also is drawing an imaginary
additional set of ways except where they are specifically marked.
--
Lester Caine - G8HFL
-----------------------------
Contact - http://lsces.co.uk/wiki/?page=contact
L.S.Caine Electronic Services - http://lsces.co.uk
EnquirySolve - http://enquirysolve.com/
Model Engineers Digital Workshop - http://medw.co.uk//
Firebird - http://www.firebirdsql.org/index.php
Roy Wallace
2009-11-30 21:10:15 UTC
Permalink
Post by Anthony
What if I map the entire section of grass which is within the right of
way as a polygon with highway=path, area=yes? ?That's how we represent
infinite overlapping criss-crossing "invisible-paths", like a
pedestrian mall.
Not bad. But what makes that area of grass a "path" as opposed to just
an area of grass you can walk on (e.g. landuse=meadow or something +
foot=yes)? Is there a difference?

I tend to think "paths" should be limited to elongated areas, designed
for or used typically for travel (other than for large vehicles like
cars), with usually a constant or slowly varying width. There's
probably a better definition though.
Anthony
2009-12-01 00:11:37 UTC
Permalink
Post by Roy Wallace
Post by Anthony
What if I map the entire section of grass which is within the right of
way as a polygon with highway=path, area=yes? ?That's how we represent
infinite overlapping criss-crossing "invisible-paths", like a
pedestrian mall.
Not bad. But what makes that area of grass a "path" as opposed to just
an area of grass you can walk on (e.g. landuse=meadow or something +
foot=yes)? Is there a difference?
Well, I didn't know landuse tags were routable. And landuse=meadow
sounds to me like a terrible tag ("meadow" is not a type of usage of
land).

But I think the key difference is that the area of land is located in
a right of way. And a second key difference is that it's useful for
routing purposes.
Post by Roy Wallace
I tend to think "paths" should be limited to elongated areas, designed
for or used typically for travel (other than for large vehicles like
cars), with usually a constant or slowly varying width. There's
probably a better definition though.
I'd say this strip of land qualifies by that definition. Length,
about 80 meters. Width: about 10-15 meters. Used quite often for
pedestrian travel (it's the way you get to the park, plus school
children regularly walk across it on their way to/from school). The
width is fairly constant.

Frankly, I don't see much point in using an area, unless you're going
to use an area for basically everything. I was kind of being
sarcastic about that. But whatever.
Martin Koppenhoefer
2009-12-01 00:21:51 UTC
Permalink
2009/12/1 Anthony <osm at inbox.org>
Post by Anthony
I'd say this strip of land qualifies by that definition. Length,
about 80 meters. Width: about 10-15 meters. Used quite often for
pedestrian travel (it's the way you get to the park, plus school
children regularly walk across it on their way to/from school). The
width is fairly constant.
what about highway=pedestrian, area=yes, surface=grass?
Post by Anthony
Frankly, I don't see much point in using an area, unless you're going
to use an area for basically everything.
in the end yes.

cheers,
Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091201/7791eee1/attachment.html>
Steve Bennett
2009-11-29 11:59:19 UTC
Permalink
Post by Roy Wallace
1) Re: connecting paths across small grass areas - don't mark a path
where there isn't one, and especially don't do it for the purpose of
trying to make routers work better. Map reality - that will always
work best in the long term. (just my personal preference)
IMHO accessible paths *must* be marked, because it's impossible to
write a router that will guess correctly. I agree that it would be
preferable not to have these hints to the router appear in the
renderer, and to be distinguished somehow. I'd almost be inclined to
invent a tag with a clearly whimsical name like highway=invisible_path
(to avoid adding to the chaos). Or even "highway=none bicycle=yes"
etc.
Post by Roy Wallace
2) Re: when to use path/footway/cycleway etc. - firstly, I prefer
highway=path because it is more extensible. Any
highway=footway/cycleway/bridleway can be expressed in terms of a
highway=path with additional access tags.
Yes. You have spelt out exactly what is wrong with this approach, with
the terms "can be expressed...with additional...tags". Succinct is
good. Semantically rich tags are good.

In this way, using
Post by Roy Wallace
highway=path can be more explicit, because of ongoing disagreements in
the definition of footway/cycleway/bridleway.
3) Re: what does <TAG> really mean? - rather than everyone giving
their personal opinion on e.g. what highway=path means, for new users
I would strongly recommend reading the wiki carefully and using that.
The wiki (in my perusal thus far) suffers from a lack of consistency
and a lack of authority. There's nowhere that says "THIS is what a
path is, and nothing else". Like Wikipedia's policies, for example.

Steve
Nick Whitelegg
2009-11-30 10:29:12 UTC
Permalink
Post by Steve Bennett
I'm doing a lot of mapping of pedestrian and bike paths around my
area, and am having trouble deciding when to use path, when footway,
and when cycleway. I'm particularly troubled by the way Potlatch
describes "path" as "unofficial path" - making it sound like an
unpaved line of footprints carved through the grass.
I don't think "highway=path" means "unofficial path". Though different
people have interpretations, highway=path seems to be used most often for
dirt paths in the countryside. An unofficial path where the landowner has
allowed access (or doesn't mind access) should be tagged as highway=path
*and* foot=permissive. Without the foot tag, many would assume the path's
private.
Post by Steve Bennett
1) In the parks near me, there are lots of paths, which I guess were
probably intended for pedestrians, but cyclists use them too.
Sometimes paved, sometimes not. I've been tagging them "highway=path,
bicycle=yes" (to be safe).
I generally use footway, rather than path, for paved paths but again this
is a contentious point.
Do you know whether bikes can access the path? If a designated bike path,
use "highway=cycleway"/"bicycle=designated" (optional). If you're not
sure, use highway=footway and leave the bicycle tag out or use
"bicycle=unknown".
Post by Steve Bennett
2) Multi-use paths, like in new housing developments. Usually paved,
and connecting streets together.
If a definite cycle path:
highway=cycleway

If not:
highway=footway; foot=permissive; [bicycle=unknown]
Post by Steve Bennett
3) Genuine multi-use paths along the sides of creeks or freeways.
Frequently with a dotted line down the middle. Most people think of
them as bike paths, but plenty of pedestrians use them too.
"highway=cycleway, foot=yes" seems the most satisfying, but according
to the definition, it should just be a "path"? I tend to assume it's a
cycleway if the gap between two entrances ever exceeds a kilometre or
so...
This would simply be highway=cycleway, I think the general assumption is
that pedestrians are permitted unless "foot=no" is added.

Nick
Steve Bennett
2009-11-30 10:39:25 UTC
Permalink
On Mon, Nov 30, 2009 at 9:29 PM, Nick Whitelegg <Nick.Whitelegg at solent.ac.uk
Post by Nick Whitelegg
Do you know whether bikes can access the path? If a designated bike path,
use "highway=cycleway"/"bicycle=designated" (optional). If you're not
sure, use highway=footway and leave the bicycle tag out or use
"bicycle=unknown".
That's a really hard question. Reflecting my biases here, but I tend to
believe I can ride my bike wherever the hell I want unless there's a sign
saying otherwise. I actually find it very to objectively decide whether
paths in my neighbourhood are "bicycle=yes". There are some narrow laneways
that I ride through - no idea if anyone else does, or whether the council
expects people to. Paths through gardens and parks are the same.

(I've noticed in the media sometimes a prevailing assumption that you can
ride a bike on a road, or on a designated bike path...and that's it. But I
think it has more to do with lack of imagination than actual restrictions.)
Post by Nick Whitelegg
Post by Steve Bennett
2) Multi-use paths, like in new housing developments. Usually paved,
and connecting streets together.
highway=cycleway
highway=footway; foot=permissive; [bicycle=unknown]
Lol. If I knew what a "definite cycle path" was, this thread wouldn't exist.
Well, I guess if there are painted bikes on the ground, it's "definite". But
that's not many.
Post by Nick Whitelegg
This would simply be highway=cycleway, I think the general assumption is
that pedestrians are permitted unless "foot=no" is added.
I wish we could codify these "general assumptions". Because they won't be
universal, which means there is bad map data being generated.

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/62c5ce50/attachment.html>
Jonathan Bennett
2009-11-30 10:58:31 UTC
Permalink
[...] I tend to
believe I can ride my bike wherever the hell I want unless there's a
sign saying otherwise.
That's fine for your personal decision making. However, for OSM we need
to provide people with as much information as possible so they can make
their own, possibly different, decisions.

Record legal access rights using access=* and bicycle=* tags, and record
physical characteristics using width=* and surface=* tags, and include
any barrier=* on the path. Routers can choose whether only to use legal
routes that way, or add to path cost where there's a bike-unfriendly
barrier in the way.

If you don't *know* the legal situation this gets tricky, but that's
something we can clear up within each country eventually.
--
Jonathan (Jonobennett)
Steve Bennett
2009-11-30 11:10:10 UTC
Permalink
On Mon, Nov 30, 2009 at 9:58 PM, Jonathan Bennett <
Post by Jonathan Bennett
[...] I tend to
believe I can ride my bike wherever the hell I want unless there's a
sign saying otherwise.
That's fine for your personal decision making. However, for OSM we need
to provide people with as much information as possible so they can make
their own, possibly different, decisions.
In case it wasn't clear, I was using the above statement to explain my
difficult in judging accurately where bikes are actually "allowed" to go.
There aren't many signs. Real observation-based tagging (surface,
smoothness, width etc...) seems like less shakey ground.
Post by Jonathan Bennett
If you don't *know* the legal situation this gets tricky, but that's
something we can clear up within each country eventually.
Ok.

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/cdcf1bc3/attachment.html>
Roy Wallace
2009-11-30 11:09:13 UTC
Permalink
Post by Steve Bennett
I wish we could codify these "general assumptions". Because they won't be
universal, which means there is bad map data being generated.
I think it's critical that this stuff be summarised on the wiki.
Besides being highly relevant to those who want to know *how to tag
things*, it might help us find a way forward out of this mess.
Steve Bennett
2009-11-30 11:45:32 UTC
Permalink
Post by Roy Wallace
I think it's critical that this stuff be summarised on the wiki.
Besides being highly relevant to those who want to know *how to tag
things*, it might help us find a way forward out of this mess.
Yep. Even if some of us don't agree that long term the "usecase vs country"
matrix is appropriate, it would be a very useful discussion point if we
could map out *current practice* this way. "Oh, the French do that???"
Post by Roy Wallace
we can have a cycleway
und einen Fahrradweg
Yep. And cycleway ~= Fahrradweg.


Steve

[originally sent to Roy only by mistake - still not used to mailing lists
that don't have reply-to list.]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/1633b056/attachment.html>
Emilie Laffray
2009-11-30 11:52:21 UTC
Permalink
2009/11/30 Steve Bennett <stevagewp at gmail.com>
Post by Steve Bennett
Post by Roy Wallace
I think it's critical that this stuff be summarised on the wiki.
Besides being highly relevant to those who want to know *how to tag
things*, it might help us find a way forward out of this mess.
Yep. Even if some of us don't agree that long term the "usecase vs country"
matrix is appropriate, it would be a very useful discussion point if we
could map out *current practice* this way. "Oh, the French do that???"
+1

Emilie Laffray
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/e7e89c30/attachment.html>
Richard Mann
2009-11-30 11:54:38 UTC
Permalink
Post by Steve Bennett
Post by Liz
we can have a cycleway
und einen Fahrradweg
Yep. And cycleway ~= Fahrradweg.
Steve
There are umpteen ways of resolving it. The problem is that we don't have a
process for agreeing which. I wouldn't go for a different highway value
personally.

Richard
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/11d25718/attachment.html>
Richard Mann
2009-11-30 10:46:23 UTC
Permalink
On Mon, Nov 30, 2009 at 10:29 AM, Nick Whitelegg <
Post by Nick Whitelegg
This would simply be highway=cycleway, I think the general assumption is
that pedestrians are permitted unless "foot=no" is added.
The crux of the matter is that this is not what the wiki says, and not what
at least some in Germany would like:

The "UK" view appears to be: foot can go anywhere (except motorways) unless
you say foot=no
The "German" view appears to be: foot can go anywhere except motorways,
cycleways and bridleways

And we have no way of resolving this :(
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/c6ca23b5/attachment.html>
Steve Bennett
2009-11-30 10:52:19 UTC
Permalink
On Mon, Nov 30, 2009 at 9:46 PM, Richard Mann <
Post by Richard Mann
The "UK" view appears to be: foot can go anywhere (except motorways) unless
you say foot=no
The "German" view appears to be: foot can go anywhere except motorways,
cycleways and bridleways
And we have no way of resolving this :(
I think you just did resolve it.

I guess the other alternative is to have some new concept of "German
bikepath" and "German bridleway", which all renderers will render the same,
but which routers will distinguish.

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/8e097a45/attachment.html>
Richard Mann
2009-11-30 10:57:31 UTC
Permalink
I didn't resolve it because either the UK view or the German view (or some
other view) has to be the default. What we can't agree is which should be
the default.
Post by Steve Bennett
On Mon, Nov 30, 2009 at 9:46 PM, Richard Mann <
Post by Richard Mann
The "UK" view appears to be: foot can go anywhere (except motorways)
unless you say foot=no
The "German" view appears to be: foot can go anywhere except motorways,
cycleways and bridleways
And we have no way of resolving this :(
I think you just did resolve it.
I guess the other alternative is to have some new concept of "German
bikepath" and "German bridleway", which all renderers will render the same,
but which routers will distinguish.
Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/f682a98c/attachment.html>
Liz
2009-11-30 11:08:08 UTC
Permalink
Post by Richard Mann
I didn't resolve it because either the UK view or the German view (or some
other view) has to be the default. What we can't agree is which should be
the default.
not at all
we can have a cycleway
und einen Fahrradweg
Roy Wallace
2009-11-30 11:11:30 UTC
Permalink
On Mon, Nov 30, 2009 at 8:57 PM, Richard Mann
Post by Richard Mann
I didn't resolve it because either the UK view or the German view (or some
other view) has to be the default. What we can't agree is which should be
the default.
Does it matter?? How hard is it to tag cycleways and bridleways with
foot=yes/no??

I would have no problem with that, if it helped give us consistency.
Steve Bennett
2009-11-30 11:14:10 UTC
Permalink
Post by Roy Wallace
Does it matter?? How hard is it to tag cycleways and bridleways with
foot=yes/no??
I would have no problem with that, if it helped give us consistency.
From a purely pragmatic perspective, the more repetitive tasks you assign to
people, the less likely it is that those tasks will be performed
consistently. I'm not convinced that telling people how to perform a task,
and getting them to do it 10,000 times will lead to 10,000 correctly
performed tasks.

(At about this stage, maybe someone should introduce some statistics into
the discussion, like number of cycleways/footways/paths tagged in various
combinations in various countries.)

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/d405e358/attachment.html>
Roy Wallace
2009-11-30 11:33:18 UTC
Permalink
Post by Roy Wallace
Post by Roy Wallace
Does it matter?? How hard is it to tag cycleways and bridleways with
foot=yes/no??
I would have no problem with that, if it helped give us consistency.
From a purely pragmatic perspective, the more repetitive tasks you assign to
people, the less likely it is that those tasks will be performed
consistently. I'm not convinced that telling people how to perform a task,
and getting them to do it 10,000 times will lead to 10,000 correctly
performed tasks.
Good point, but I think it's ok to first work out how we *should* be
tagging, before we assume that people will stuff it up.
Martin Koppenhoefer
2009-11-30 11:31:09 UTC
Permalink
2009/11/30 Richard Mann <richard.mann.westoxford at googlemail.com>
Post by Richard Mann
On Mon, Nov 30, 2009 at 10:29 AM, Nick Whitelegg <
Post by Nick Whitelegg
This would simply be highway=cycleway, I think the general assumption is
that pedestrians are permitted unless "foot=no" is added.
The crux of the matter is that this is not what the wiki says, and not
The "UK" view appears to be: foot can go anywhere (except motorways) unless
you say foot=no
The "German" view appears to be: foot can go anywhere except motorways,
cycleways and bridleways
And we have no way of resolving this :(
there is some ways to resolve this:
- use a polygon (border) to determine whether the path is in Germany or in
the UK
- explicitly tag foot=yes in the UK or foot=no in Germany on those ways (or
both for best consistency).
- use path like described in the wiki and tag according to the right of ways
and tags proposed


cheers,
Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20091130/6b4e1317/attachment.html>
Loading...