[OSM-talk-be] Path vs Footpath (sorry for opening the pandora box)
Francois Gerin
francois.gerin at gmail.com
Wed Feb 24 08:13:25 UTC 2021
Thank you Steven, finally something in the good direction!
Fixing the issue is clearly the thing to do and much better than adding
a new mess on top of the existing one, which would necessarily make any
real fix impossible.
But... (yes) If one has the energy to start this process, here are a few
hints he should keep in mind at any time:
1. It's important to really understand the problem before trying to fix
it. And here, it looks like few people understand where+what it is. See
tag vs attribute here below.
2. Rules are important for community projects, especially wide ones. OSM
is a huge and specific one. But beyond respecting rules, there is
something at least as much important: the respect of
hierarchy/structure. One should understand the distinction between the
national OSM branch and the global/top level.
If every national branch fixes a global issue at its level, first,
things are quickly going to be incompatible between countries, second,
it makes it much more complicated, if not impossible, to correct later
on at the top/global level.
Fixing things globally is necessary to make the project survive on the
long term, for the same reasons.
3. Here after, I will use the term *walker_way* in place of "footway",
because the term itself is misleading and readers stick with the human
meaning, thus they miss the technical aspect which has first priority.
If you see clear in the tag vs attribute, key/value, DB id, etc. You'll
see these are just aliases.
Those familiar with XML/HTML know that tags and attributes are not the
same, even if sometime a problem may be addressed locally with either.
Similarly, people used to OOP (Object Oriented Programming) know the
difference between an object and its characteristics
(members/attributes). That's the same with keys/values in the context of
OSM: the different values for highway point to different
entities/categories. There is NO reason to suddenly treat one category
via attributes while all others are treated appropriately. This is not
for rendering, this is for sane data into the database.
Are *walker_way* a useful category? - Yes, even if it looks like no at
first glance. (Story: I also was puzzled when facing this questioning
myself, the first time. I changed and changed again according to my
understanding of the wiki... To, finally, lately, understand the meaning
and reasons it exists.)
=> This category exists, it is useful, even if it applies only to a
fraction of cases. The mistake/error/problem is neither that this
category exist, nor even its misleading name, but the documentation
around. Specifically the fact the explanations, definitions, usages,
recommendations are spread in multiples locations, sometime with a look
of contradictions.
NOTE: See the wood of Marche-les-Dames. There are tracks that switch to
paths. That switch is not easy to locate, they move between seasons,
they are particularly subjective, because of the width/aspect/personal
interpretation/consideration. However, not far from there, there are
*walker_ways* that are definitely something different than a path or a
track, and for which the definitions and rules of signaling (roadsigns)
do not apply, being for bikes, horses or anything else. But they are
clearly "_*mainly for walkers*_", as stated in the wiki pages.
I went across the documentation again and again, to end with the
conclusion that, NO, the current structure is NOT bad, there is no error
there, it was the right structure and *walker_ways* fully apply, are
useful. No signaling or width or the likes (attributes) can replace that
category, without breaking the good structure.
4. The real issue is the documentation, the lack of respect of the
hierarchy (global vs national) because of an initial slightly messy
wiki. And now, the documentation drifts further, never-ending drafts
take more and more precedence on official decisions/consensus/rules
(good or bad), to the point people do not respect the global ruling that
is mandatory to maintain the project on the long term. We are soon going
to face multiple national OSM in place of a worldwide map.
Worse, that approach kills any possibility to maintain the project for
future generations of mappers/maintainers, because it won't be reversed.
Spaghetti code... you know.
5. To write it again with other words, the issue is not the term
"footway". The issue is that it is interpreted with a human meaning
while it is an ID in a database... Just there to identify a specific
category. The category is useful, the structure is good, please don't
change this. Rename, via an alias or so, and fix the documentation, at
the global level, where stands the root cause.
When the global level will get fixed, and only then, a national
addition/supplement/improvement could be considered. But you will notice
that it is no more necessary, the problem would be fixed for all
countries at once.
=> National rules should just be supplements/additions based on the
global ruling. Only for country-specific subtleties. Not respecting this
kills the maintainability/lifetime expectation of the project.
6. If the conclusion ends with a change of the keyword "footway", in
order to stop new mappers to reproduce the same mistakes, then an alias
has much more chances to be accepted at the global level, since its
implementation is quite immediate while any other kind of change would
be very difficult, hence probably rejected. If not rejected, the process
would anyway take at least 2 years, probably more. With an alias, a
"solution" could be implemented in just a few weeks, in a quite
transparent way.
If such an alias is chosen, then consider a keyword that makes clear it
is a technical ID, not a human-meaning word from the dictionary. Look at
the *walker_way* keyword that I invented here above... Look at what the
underscore brings... That looks like a "technical" keyword. Not a
dictionary word to be interpreted differently in a subjective way, just
an ID.
I hope it will help you to recover things and bring back the project in
a good, long term direction.
Regards,
François
On 23/02/21 15:54, Steven Clays wrote:
> In the UK one cannot ride a bicycle on a public footpath. Very often,
> it is physically impossible.
> I think we are making each other angry for almost no reasons. We wight
> even consider refining the rules, but I agree with François that he
> follows official rules. The disputed paths only concern a very minimum
> of the whole lot. So, François, your time spent is not to be
> considered as wasted!
> I used to tag everything with footway, nowadays I only use it for
> exclusive pedestrian use. Slow roads in Belgium can be used (even when
> they are not referenced!) for all types of users that can use the
> width of the path and that are not excluded by roadsigns. So,
> technically, a footway does not apply.
> We might even consider to put the slowroads Belgium page to a vote or
> how does this process work?
>
>
>
>
>
> Op ma 22 feb. 2021 om 13:55 schreef Marc Gemis <marc.gemis at gmail.com
> <mailto:marc.gemis at gmail.com>>:
>
> Originally, the highway-tags comes from the UK. So
> highway=footpath really means a small road designated for
> pedestrians. They are signposted like that. I'm not really sure,
> but I think that this classification does not mean that you cannot
> ride on it with a bicycle.
>
> The Germans and others didn't like this, as they do not know roads
> that are footpaths, so they introduced highway=path.
> And then things became very messy. And then the standard renderer
> started showing things in the same way.
>
> I think many people say highway=footway and highway=path,
> foot=designated (bicycle =no) are synonymous. Surface, nor
> location play a role.
>
> Some people as HW=footway has to be paved, and in a town, etc.,
> but this all was added later and is afaik not accepted all over
> the world. Sidewalks are not paved in all countries, hence it is
> not a good idea to require that.
>
> I think there is no general consensus, not in Belgium, not in the
> rest of the world on whether once should use path or footway. For
> me, HW= footway is just a shortcut for HW=path +
> foot=designated,bicycle=no.
>
> regards
>
> m.
>
>
> On Mon, Feb 22, 2021 at 9:34 AM Francois Gerin
> <francois.gerin at gmail.com <mailto:francois.gerin at gmail.com>> wrote:
>
> I'm sorry, it's really nice that each one interprets things
> like he wants, but official rules are there to avoid issues,
> especially in subtle cases (like current case, as already
> mentioned) and to avoid individual _*interpretations*_.
>
> In the meantime, I saw a lot of inconsistencies in the
> previous exchanges, while you definitely did not take into
> account the elements I shared, which are based and, AGAIN,
> have been confronted to the official rules.
>
> Read again the main definitions
> <https://wiki.openstreetmap.org/wiki/Key:highway#Paths>, they
> are CLEAR. Especially the usage of a path and recommandations.
>
> You don't like the use of footways, that's really sad, but I
> RESPECT THE RULES. A path is NOT to use in the cases I
> mentioned, like confirmed by the usage rules that are
> OFFICIAL, link here above.
>
>
> Contradictions (non-exhaustive list):
>
> - You claim it's cheating on rules for rendering purpose. But
> you recognize the rendering is not different between a path
> and a footway.
>
> - You rely on a never-ending draft, while official rules have
> precedence and are CLEAR. The draft itself, which has not come
> to an official rule change for years and introduce some
> confusions (this one, among others), which you make use of to
> torn things in the way you would like things to be. But rules
> are rules and are there for good reasons. When you tell "must
> be understood as", "interpret like" or the likes, you are
> directly demonstrating the lack of understanding of why rules
> exist.
>
> - You claim a path must be used, while the official
> definition of a path, and the recommended usage CLEARLY tells
> it's a footway.
>
> - You make distinction between roads (secondary vs
> residential, motorway vs secondary) and track vs path, but you
> cannot make a distinction between a path and a footway, which
> is clearly a bigger/clearer gap, much clearer than track vs
> path, which are often difficult to distinct for on site.
>
> - You "don't see why", so it does not exist. Go to sites on
> see! Rules are there for good reasons. NOTE: When someone
> insist on the fact something is subtle and complex, rejecting
> without even considering the case/reason is error prone.
>
>
> Particularly sad to see how OSM is turning against itself,
> like often with community projects. After having spent wasted
> hundreds of hours to clean the mess it was in many parts, in
> RESPECT OF THE RULES even when I didn't like them, I come to
> the conclusion it was just a wast of time. => My contributions
> to OSM stops here.
>
>
>
>
> On 20/02/21 18:24, ghia wrote:
>>
>> I agree!
>>
>>
>> There seems at first sight no raison to differentiate the
>> paths and to map some as footway, and others as path. I think
>> they should all be mapped as path.
>>
>> If there are some with bicycles allowed, this can be tagged
>> by bicycle yes or no.
>>
>> Regards,
>>
>> Gerard
>>
>>
>> On 2021-02-19 15:06, Matthieu Gaillet wrote:
>>
>> First, I would certainly not break a good work - even if
>> I disagree - like you did here for example:
>> https://www.openstreetmap.org/#map=16/50.4822/4.9482
>> <https://www.openstreetmap.org/#map=16/50.4822/4.9482>. That's
>> clearly a work done with precision and evidently the tags
>> are used consistently.
>> That said, I'm currently checking / correcting all the
>> uses of the footway tag outside urbanised areas in the
>> Haute-Meuse region (from Givet to Namur) and in 99,99% of
>> cases, at least from the consistency point of view, it
>> was an error of the mapper, mostly influenced by politics
>> (mapper that wants to emphasis the pedestrian character
>> of a vicinal path) or simply by mistake or ignorance.
>> Actually I don't really care about the use of the footway
>> tag like you did IF and only IF it is consistent all over
>> OSM and that there is a consensus about that use. As far
>> as I can see your way of doing things is an isolated case.
>> I really believe that there is a misunderstanding in the
>> definition of the _word_ itself. Have a look at that
>> query of google Images
>> <https://www.google.com/search?q=footway&client=safari&hl=fr&source=lnms&tbm=isch&sa=X&ved=2ahUKEwjl4pmUgfbuAhVMwKQKHXTgDnQQ_AUoAnoECAgQBA&biw=1280&bih=642> on
>> the word footway. Even if those footways can be unpaved
>> (clay for example), all of the pictures refers to
>> _urbanised places_. The word footway refers to
>> "trottoirs" or "voetpad" much more than "sentier" or "pad".
>> Therefore, with all due respect to your work, I believe
>> that you're wrong 😊 It might be considered as mapping
>> for the renderer since default OSM maps don't really make
>> the difference between a "normal" path and a path tagged
>> as narrow or difficult. But even if your approach could
>> make sense, the use of that footway tag is wrong for me.
>> Other renderers are perfectly using the trail_difficulty
>> and trail_visibility tags that are made for such use.
>> On the screen capure of opentopomap
>> <https://opentopomap.org/#map=16/50.31757/4.82489> below,
>> the green arrow shows a "normal" path, where the red one
>> shows a difficult path I mapped recently. It works !
>> Again and finally : for the immense majority of the
>> people, a "footway" is a _safe place to walk_.
>> Definitively not an alpine path.
>> Matthieu
>>
>> On 18 Feb 2021, at 13:32, Francois Gerin
>> <francois.gerin at gmail.com
>> <mailto:francois.gerin at gmail.com>> wrote:
>>
>> Hi all,
>>
>> I faced the same question a while ago. I also
>> realized the lack of consensus, but also the good
>> reason for the lack of consensus: the problem is not
>> that simple, and there are different points of view,
>> sometime very opposite, but also with a good common base.
>>
>> First, for years I didn't changed the map when
>> already mapped. And I mapped "like the area around"
>> to be consistent. But then I became more active and
>> started to refresh and map not-yet-covered areas in
>> my region. Mainly woods and forest, where less
>> mappers work and because there was a clear need.
>> => Quickly I realized that it was really important to
>> map "appropriately" in such areas. And, just for the
>> confirmation, I encountered several times people
>> lost, sometime with babies and bikes in quite
>> dangerous areas. (That happened many times in the
>> forest of Marche-les-Dames, that a full refreshed
>> recently.)
>>
>> I ended with this "simple" approach, which is also
>> the best "consensus" (for myself) from the different
>> definitions, remarks, wiki pages, and compatible with
>> forest use:
>>
>> - Everything is mostly a path, except if it is a
>> track or a footway.
>>
>> - A track is where a 4-wheels vehicle, more
>> specifically a forestry tractor, can (if traces on
>> the ground) or is used to pass. (Distinction made so
>> that a path does not become a track just because a
>> quad can pass!)
>>
>> - A footway is definitely useful: this is a path too
>> small for horses and mountain bikes. (By mountain
>> bikers, I mean "standard people", aka end users, not
>> pro mountain bikers who can pass nearly everywhere a
>> pedestrian passes!) That definitely correspond to
>> what bikers call "singles": a very small track, where
>> two bikes cannot pass side by side.
>>
>>
>> When I reached that approach, I read again the
>> different points of view, remarks, wiki, to conclude
>> that it was respecting quite well most of the points
>> considered important. And, for me, it also satisfied
>> the need to make that distinction, which is important
>> on site. (cf. lost people, dangerous situations)
>>
>> Note that even if I'm a biker, I force myself to
>> consider OSM for the "end user". With the distinction
>> I make for a "standard end user": Consider the map
>> like a family getting out for a walk on the Sunday...
>> Neither for a pro mountain biker, nor a horse driver.
>> Even if those categories probably benefit the most
>> from the distinction.
>>
>> My 2 cents.
>>
>> ++
>> François
>>
>>
>>
>> On 18/02/21 12:49, Pieter Vander Vennet wrote:
>>
>> Hi everyone,
>>
>> Sorry to send a second email just after my first,
>> but while doing some more research, I found that
>> this controversy is already pretty old; see:
>>
>> https://www.openstreetmap.org/user/Richard/diary/20333
>> <https://www.openstreetmap.org/user/Richard/diary/20333>
>>
>> and
>>
>> https://wiki.openstreetmap.org/wiki/Path_controversy
>> <https://wiki.openstreetmap.org/wiki/Path_controversy>
>>
>> Kind regards,
>> Pieter
>>
>> On 18.02.21 12:34, Pieter Vander Vennet wrote:
>>
>> Hi Matthieu,
>>
>> Welcome in the swampy fields of tagging
>> discussions ;)
>>
>> *My view*
>>
>> First of all, we do professional
>> routeplanning, for both cyclists and
>> pedestrians. And yes, I do (mostly) agree
>> with your view: a path is a small, unpaved
>> (desire) path, e.g. through a forest whereas
>> a footway is IMHO a typical paved (or
>> planned) road of at least 0.5m wide. A rule
>> of thumb that I use is that a
>> wheelchair/stroller could pass easily, or as
>> Gerard said earlier: "it is like a sidewalk,
>> but just not next to a road"
>>
>> If the "footway" is sufficiently wide that a
>> car /could/ drive over it (but is not allowed
>> to), I'm inclined to mark them as
>> /highway=pedestrian/. This is useful
>> information, as e.g. emergency services might
>> take it during an intervention to get close
>> to the location of the accident.
>>
>> I'm also inclined to mark a wide, planned way
>> (e.g. in parks) as footways too.
>>
>> I try to base my road classification mostly
>> on physical aspects: a path stays a path,
>> even if it suddenly has a name board. This is
>> because of my view from routeplanning: in
>> general, I assume that that a footway is
>> accessible to a wheelchair user, whereas a
>> path is not. To explicitly add the vicinal
>> road status, there are some tags for that
>> (vicinal_road:ref IIRC?). This is the only
>> place where I disagree with you:
>>
>> > The only exception I see is a path in the
>> country side that is explicitly marked (road
>> signs) as pedestrian only, and/or has
>> turnstiles or other gates to keep other users
>> away.
>> I would still mark those as a `highway=path`,
>> with an additional `bicycle=no` and map the
>> turnstiles/kissing gates explicitly. The data
>> consumer can then decide what to do.
>> Note however that not everyone agrees with my
>> vision and that I'm not always consistent too
>> - I mapped a very peculiar case
>> <https://www.openstreetmap.org/way/107877794>
>> yesterday that by my objective criteria
>> should be a 'path', but that I mapped as
>> footways due to their context as that felt
>> more appropriate - butthat place
>> <https://www.openstreetmap.org/way/126568080>
>> has given me more tagging questions too...
>>
>> At last, some people say that "a footway
>> needs a traffic sign to be a footway" or "a
>> cycle path needs a traffic sign to be a
>> cyclepath". That is a view I vehemently
>> reject - not every qualitive footway has a
>> traffic sign nor has every traffic sign a
>> qualitative footway - although a traffic sign
>> can help in making these decisions.
>>
>> Also abusing `highway=path` for shared
>> infrastructure cycle/pedestrian
>> infrastructure is something I loathe: it
>> erases a lot of information and is an
>> effective downgrade of the relevant ways from
>> a routeplanning perspective, as we have to
>> assume the way is a desire path (small,
>> unpaved); not accessible to e.g. wheelchairs,
>> strollers and rollerskate, instead of the
>> very accessible nicely paved, wide footway.
>> To be able to replicate all the information
>> for this downgrade, we would need
>> `surface=*`, `width=*`, `smoothness=*` and
>> maybe even `wheelchair=*` to be sure it is a
>> highly qualitative footway and quite a bit of
>> tricky and inexact preprocessing. However, I
>> do not have a perfect solution for the shared
>> footways/cycleways as well - but marking as
>> path is definitively worse.
>> So, Marc_marc: I'm sorry, but I do not agree
>> with you and some of the wiki definitions!
>> But that is fine - a disagreement is often
>> due to a different perspective or some
>> missing information. And OSM won't fail over
>> a bit of disagreement ;)
>>
>>
>> *Some history*
>>
>> Apart from my vision, it is also important to
>> know that OpenStreetMap started in the UK,
>> where there are plenty of vicinal roads. I
>> think those where historically mapped as
>> highway=footway too, but I'm not sure of
>> that. Furthermore, as Gerard nicely stated
>> earlier, it is a common translation error.
>>
>> Furhtermore, the iD editor used to "upgrade"
>> tags: a `highway=footway + bicycle=yes` and
>> `highway=cycleway + foot=yes` got upgraded to
>> `highway=path; bicycle=yes; foot=yes`. As
>> the iD editor is widely used, there are quite
>> some footways downgraded now...
>>
>> Kind regards,
>> Pieter
>>
>> On 18.02.21 10:27, Matthieu Gaillet wrote:
>>
>> Hi,
>> I would like to know if there is some
>> kind of consensus in Belgium regarding
>> the use of <footway> and <path> tags.
>> My intuitive interpretation in the
>> following :
>>
>> * a footway, generally speaking, is
>> anything that is specifically created
>> for pedestrians in urbanised areas.
>> * a path, is generally speaking
>> anything that is not a track (thus
>> not for 4 wheeled vehicles) and not
>> (as well) paved like a footway.
>>
>> I know there are other much more loose
>> interpretations that say that a footway
>> might be a non-paved path, but my
>> question is : why would one tag them
>> differently than others ? After all, a
>> path is not suitable for anything else
>> than pedestrian use (except sometimes
>> bikes) ? On the contrary, footways in
>> urbanised places *are* special and it
>> makes sense to map them differently.
>> I observe that some mappers are using the
>> footway tags for paths in forests or
>> fields in the middle of nowhere. Those
>> are often "sentiers communaux" (public
>> paths) mapped by balnam affiliates. Its
>> driving me nuts 😊
>> - most of the time this difference in the
>> way those paths are mapped doesn't
>> reflect any physical nor practical
>> reality on the field.
>> - this creates vagueness and looseness, I
>> see "normal" paths suddenly showed as
>> "special" on maps without any clear reason.
>> - some could argument that the path tag
>> is not detailed enough. That's not true :
>> it can be (and is) combined with a lot of
>> other tags to qualify it from multiple
>> point of views and renderers are already
>> taking care of them. This is *not* the
>> case of the footway which is (logically)
>> kind of monolithic.
>> The only exception I see is a path in the
>> country side that is explicitly marked
>> (road signs) as pedestrian only, and/or
>> has turnstiles or other gates to keep
>> other users away.
>> Do you generally agree with my way of
>> seeing things ? Is it at least the
>> general way of doing things in Belgium ?
>> Thanks for sharing your thoughts.
>> Matthieu
>>
>> _______________________________________________
>> Talk-be mailing list
>> Talk-be at openstreetmap.org <mailto:Talk-be at openstreetmap.org>
>> https://lists.openstreetmap.org/listinfo/talk-be <https://lists.openstreetmap.org/listinfo/talk-be>
>>
>> --
>> Met vriendelijke groeten,
>> Pieter Vander Vennet
>>
>> --
>> Met vriendelijke groeten,
>> Pieter Vander Vennet
>>
>>
>> _______________________________________________
>> Talk-be mailing list
>> Talk-be at openstreetmap.org <mailto:Talk-be at openstreetmap.org>
>> https://lists.openstreetmap.org/listinfo/talk-be <https://lists.openstreetmap.org/listinfo/talk-be>
>>
>> _______________________________________________
>> Talk-be mailing list
>> Talk-be at openstreetmap.org
>> <mailto:Talk-be at openstreetmap.org>
>> https://lists.openstreetmap.org/listinfo/talk-be
>> <https://lists.openstreetmap.org/listinfo/talk-be>
>>
>>
>> _______________________________________________
>> Talk-be mailing list
>> Talk-be at openstreetmap.org <mailto:Talk-be at openstreetmap.org>
>> https://lists.openstreetmap.org/listinfo/talk-be
>> <https://lists.openstreetmap.org/listinfo/talk-be>
>>
>>
>> _______________________________________________
>> Talk-be mailing list
>> Talk-be at openstreetmap.org <mailto:Talk-be at openstreetmap.org>
>> https://lists.openstreetmap.org/listinfo/talk-be <https://lists.openstreetmap.org/listinfo/talk-be>
> _______________________________________________
> Talk-be mailing list
> Talk-be at openstreetmap.org <mailto:Talk-be at openstreetmap.org>
> https://lists.openstreetmap.org/listinfo/talk-be
> <https://lists.openstreetmap.org/listinfo/talk-be>
>
> _______________________________________________
> Talk-be mailing list
> Talk-be at openstreetmap.org <mailto:Talk-be at openstreetmap.org>
> https://lists.openstreetmap.org/listinfo/talk-be
> <https://lists.openstreetmap.org/listinfo/talk-be>
>
>
> _______________________________________________
> Talk-be mailing list
> Talk-be at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-be/attachments/20210224/66c3e5b7/attachment-0001.htm>
More information about the Talk-be
mailing list