Branched and alternative roujtes

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

Branched and alternative roujtes

Kevin Kenny-3
(Summary: What do the data *consumers* want to see in the tagging for
route alternatives, circular routes, and routes that begin and end on
dual carriageways?)

On Mon, Aug 19, 2019 at 3:47 AM Sarah Hoffmann <[hidden email]> wrote:
> We do happen to have a clear rule for unbroken linear routes: just assemble
> in the obvious way, no matter if sorted or unsorted. We don't have any rule
> for anything more complicated that mappers can follow to get the desired
> effect. We already fail with something as simple as a directed unbroken
> linear routes and circular routes. There is no single recommended way to
> define the start point.

A circular route may not even have a start point.  Hikers doing the
Carnberry Lake 50 can start and end it anywhere and do it clockwise or
anti-clockwise (although there happen to be only a couple of good
places to get on and off the circular route).

> Assuming we don't care what happens to really botched relations, all cases
> except one that I listed initially are covered with one single simple
> instruction to the mapper: sort your route.
>
> What remains are routes which are split/have alternatives/access routes etc.
> Gut feeling tells that roles will solve those cases but I get back to you
> on that once I had a go at implementing it.

If I recall correctly, you're well on the way to handling role=forward
and role=backward splits - I seem to recall that WMT was fairly
graceful about the one or two of those I put in (to repair routes that
were broken altogether - I'm not in the habit of editing cycling
routes otherwise).

For hiking routes, the splits that I have and don't quite know how to
manage are:

- Short diversions. Some of the trails that I've mapped have short
segments for winter and summer routes, or marked alternatives for use
in case of high water, fire-season trail closures, or beaver activitty
(which, I suppose is a special case of high water).

- Access ways. Ordinarily these are marked separately from the main
trail and I just carry them as separate routes. The only case where
I've really wanted to make some sort of association is that the Green
Mountain Club, in addition to the 'end to end' award on the Long Trail
(for hiking the main stem) offers a much more difficult 'side to side'
award for hiking each of the approach trails. I have Absolutely No
Idea how to represent this, if I were to do so.

- Major diversions. For the 'end to end' award on New York's Long
Path, the suburban sections in Orange County are recognized to be a
problem for hikers, and a recognized alternative is to leave the Long
Path in Harriman Park, follow the Appalachian Trail to High Point, New
Jersey, and then the Shawangunk Ridge Trail to rejoin the Long Path in
Otisville.  I'm perfectly fine, though, with simply offering this as
narrative, and having the relations show this as three separate
trails. Hikers have to make their own decisions sometime!

- Trails that are waymarked only in one direction. I do this with
'oneway=yes' on the relation, and order the ways accordingly, but I
did encounter a circular route that seemed to be ambiguous however I
did it. (The trail maintainers rendered this one moot by installing
signs facing the other way.)

The same sort of things seem to infect road routes:

- Routes that begin and end at the interchanges among dual
carriageways, which give no single point that can be indentified as an
endpoint because the geographic endpoints are different in the two
directions. JOSM has a real problem sorting these.

- Multiple-carriageway routes, where there are grade-separated ways
between 'express' and 'local', or between 'auto' and 'hgv' or between
'vehicles with transponders' and 'vehicles paying cash'.  This is an
additional layer of split on top of 'forward' and 'backward' and is
even worse for messing up sorting.

- The same thing can happen on surface streets where there are
numbered routes that are bannered 'ALT, 'BUSINESS', 'TRUCK', etc. For
these, though, I'm perfectly fine with saying that the loops and spurs
are entirely separate routes. The signage is distinct, and people in
the affected areas are used to being directed onto 'US 20 Business' or
whatever.

We have 'forward' and 'backward' pretty much conquered (except for the
dual-carriageway case). WMT already appears to figure it out, (well,
mostly), and JOSM successfully sorts these, even when routes traverse
roundabouts.

I'd like to hear from data consumers in particular what tagging they'd
like to see for

- diversions and alternatives
- routes with different endpoints in the forward/backward directions
- spur routes
- one-way routes that may be circular

[End of technical discussion. Political jeremiad follows. Feel free to ignore.]

With my data consumer hat on, I want to be able to consume the data
describing these situations without having to guess, because computers
are very, very bad at guesswork.

With my mapper hat on (the current one is a faded and stained baseball
cap bearing the logo of a trail crew), I want to produce data that
will be useful to consumers. I am told that there are other mappers
who are more 'artists' interested in describing the world around them
without regard to utility, but I think that most mappers would agree
with me in wanting the data to be useful to others.

That's why I'm interested in hearing specificallly from data consumers
- just as mappers have the actual knowledge of what's in the world,
consumers have the actual knowledge of how to aggregate it and present
it to others in a useful fashion.

I'm willing to conform with reasonable mapping guidelines. Where I've
objected in the past to specific mapping guidelines, it's because of
one of these sorts of things occurred:
 - the guideline amounted to 'don't map that!'
 - the guideline was self-contradictory
 - the guideline provided me no way to distinguish features that my
rendering needed to distinguish. (That is NOT 'tagging for the
renderer,' guys!)
 - the guideline provided me no way to 'rough out' features that I'd
observed in the field without needing to return for more observations.
(Or worse, provided no way to map the object without doing research at
home because information needed for conformance was not observable in
the field.)

For many 'it's too much work for the mapper' objections, the solution
is 'make the editor smarter', and the maintainers of the editors have
been quite good about responding to tagging schemata that actually
solve a problem and come into common use.

--
73 de ke9tv/2, Kevin

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Branched and alternative roujtes

Paul Allen
On Mon, 19 Aug 2019 at 15:13, Kevin Kenny <[hidden email]> wrote:
(Summary: What do the data *consumers* want to see in the tagging for
route alternatives, circular routes, and routes that begin and end on
dual carriageways?)

Since you've broadened the discussion to deal with more than just walking/
hiking/cycling routes, I'll take the opportunity to mention a bus route (yes,
THAT one, yet again).  It may seem somewhat irrelevant to walking routes
as it has complications they (usually) do not, but a solution that handles
this bus route well would probably be useful for other bus routes and other
types of route.

It's a very messy circular route.  I don't see how any algorithm could correctly
figure out the actual sequence of ways traversed unless they were sorted
correctly.  Three times it goes into a cul-de-sac, reverses into a side-road
(that is also a cul-de-sac) then goes forward in the opposite direction from
whence it came.  In one place it does the same reverse-turn trick in the middle
of a very long road because it doesn't go all the way.  In another place it goes around
four sides of a square, looping the loop.  In one place it traverses the same sequence
of ways twice, about 30 minutes apart.  Even the drivers occasionally get it

Even with a close inspection of the one-way streets along the route, it's
impossible to figure out exactly the sequence in which it traverses the route.
And yet the information is there (I hope, but there may be errors) in the ordering
of the relation.  We don't seem to have a tool that would let an ordinary user
figure it out easily, but a user could (with a great deal of time and effort)
use the query tool of standard carto to get the route, then work his or way through
the list of ways in the route by clicking on them in turn, then returning back to the list
each time. A slightly more savvy user would right-click on each way in turn to open
it in a new tab, but it's still a lot of time and effort.

At this point I had a thought.  Given what we already have in standard carto's query
tool, it would be a Simple Matter Of Programming[tm] to add a way of dealing with
routes.  When I say "SMOP" it could be anything from an hour of trivial coding to
weeks and weeks of a complete rewrite, but that's just a matter of details and
some Dunning-Krugeresque hand waving on my part.

The way the query tool works is to return a list of nearby objects.  Hover over any
object in the list and it is highlighted in a browny-orange.  Very useful.  Suppose
that sort of highlighting also worked with the list of ways in a route relation (as in
the link above).  The whole route is highlighted in browny-orange.  But if hovering
over a way in the list caused that way to be highlighted in a different colour, you
could easily see the steps in the route and the sequence in which they are
traversed (assuming it was correctly sorted, of course), by hovering your way through
the list.

Things get complicated with alternate routes and variant routes, but I'll just
do some more Dunning-Krugeresque hand-waving here.

Of course, we're always going to have routes that aren't sorted.  Partly because
some editors disordered routes (they seem not to do so these days, although it's
possible they get confused by rare cases).  Partly because some mappers don't
realize they should do so (although mappers would tend to add the ways of a route
in sequence and a good editor would maintain that sequence).  Partly because some
mappers think it's their mother's job to tidy their bedroom (sorry, I meant the router's
job to make sense of what they've mapped).  I suspect that if standard carto's query
permitted routes to be inspected that way, more mappers would take care to ensure
their routes were sorted.

--
Paul


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Branched and alternative roujtes

Richard Fairhurst
In reply to this post by Kevin Kenny-3
My use-case for cycle.travel is having a single polyline that I can make into
a route guide at https://cycle.travel/routes . Currently there’s two dozen:
I’d like there to be thousands. So:

> - diversions and alternatives

Give them consistent roles so I can ignore them.

> - routes with different endpoints in the forward/backward directions

Not fussed. I only do the route in one direction.

> - spur routes

Again, consistent roles so I can filter them out.

> - one-way routes that may be circular

If there’s an agreed start point, then put the node in the relation with an
appropriate role.

cheers
Richard



--
Sent from: http://gis.19327.n8.nabble.com/Tagging-f5258744.html

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Branched and alternative roujtes

Warin
In reply to this post by Kevin Kenny-3
On 20/08/19 00:11, Kevin Kenny wrote:

> (Summary: What do the data *consumers* want to see in the tagging for
> route alternatives, circular routes, and routes that begin and end on
> dual carriageways?)
>
> On Mon, Aug 19, 2019 at 3:47 AM Sarah Hoffmann <[hidden email]> wrote:
>> We do happen to have a clear rule for unbroken linear routes: just assemble
>> in the obvious way, no matter if sorted or unsorted. We don't have any rule
>> for anything more complicated that mappers can follow to get the desired
>> effect. We already fail with something as simple as a directed unbroken
>> linear routes and circular routes. There is no single recommended way to
>> define the start point.
> A circular route may not even have a start point.  Hikers doing the
> Carnberry Lake 50 can start and end it anywhere and do it clockwise or
> anti-clockwise (although there happen to be only a couple of good
> places to get on and off the circular route).
>
>> Assuming we don't care what happens to really botched relations, all cases
>> except one that I listed initially are covered with one single simple
>> instruction to the mapper: sort your route.
>>
>> What remains are routes which are split/have alternatives/access routes etc.
>> Gut feeling tells that roles will solve those cases but I get back to you
>> on that once I had a go at implementing it.
> If I recall correctly, you're well on the way to handling role=forward
> and role=backward splits - I seem to recall that WMT was fairly
> graceful about the one or two of those I put in (to repair routes that
> were broken altogether - I'm not in the habit of editing cycling
> routes otherwise).
>
> For hiking routes, the splits that I have and don't quite know how to
> manage are:
>
> - Short diversions. Some of the trails that I've mapped have short
> segments for winter and summer routes, or marked alternatives for use
> in case of high water, fire-season trail closures, or beaver activitty
> (which, I suppose is a special case of high water).
>
> - Access ways. Ordinarily these are marked separately from the main
> trail and I just carry them as separate routes. The only case where
> I've really wanted to make some sort of association is that the Green
> Mountain Club, in addition to the 'end to end' award on the Long Trail
> (for hiking the main stem) offers a much more difficult 'side to side'
> award for hiking each of the approach trails. I have Absolutely No
> Idea how to represent this, if I were to do so.
>
> - Major diversions. For the 'end to end' award on New York's Long
> Path, the suburban sections in Orange County are recognized to be a
> problem for hikers, and a recognized alternative is to leave the Long
> Path in Harriman Park, follow the Appalachian Trail to High Point, New
> Jersey, and then the Shawangunk Ridge Trail to rejoin the Long Path in
> Otisville.  I'm perfectly fine, though, with simply offering this as
> narrative, and having the relations show this as three separate
> trails. Hikers have to make their own decisions sometime!
>
> - Trails that are waymarked only in one direction. I do this with
> 'oneway=yes' on the relation, and order the ways accordingly, but I
> did encounter a circular route that seemed to be ambiguous however I
> did it. (The trail maintainers rendered this one moot by installing
> signs facing the other way.)
>
> The same sort of things seem to infect road routes:
>
> - Routes that begin and end at the interchanges among dual
> carriageways, which give no single point that can be indentified as an
> endpoint because the geographic endpoints are different in the two
> directions. JOSM has a real problem sorting these.
>
> - Multiple-carriageway routes, where there are grade-separated ways
> between 'express' and 'local', or between 'auto' and 'hgv' or between
> 'vehicles with transponders' and 'vehicles paying cash'.  This is an
> additional layer of split on top of 'forward' and 'backward' and is
> even worse for messing up sorting.
>
> - The same thing can happen on surface streets where there are
> numbered routes that are bannered 'ALT, 'BUSINESS', 'TRUCK', etc. For
> these, though, I'm perfectly fine with saying that the loops and spurs
> are entirely separate routes. The signage is distinct, and people in
> the affected areas are used to being directed onto 'US 20 Business' or
> whatever.
>
> We have 'forward' and 'backward' pretty much conquered (except for the
> dual-carriageway case). WMT already appears to figure it out, (well,
> mostly), and JOSM successfully sorts these, even when routes traverse
> roundabouts.
>
> I'd like to hear from data consumers in particular what tagging they'd
> like to see for
>
> - diversions and alternatives
> - routes with different endpoints in the forward/backward directions
> - spur routes
> - one-way routes that may be circular

There needs to be roles for these (primarily a mapper here).

Are not diversions and alternatives the same thing? Or at least close enough that they can use the same OSM tags?

I see spurs differently from access tracks. Spurs may be dead ends off to a look out or camp site.

Access tracks provide access to the actual route, they may come from a train station or other easily accessed point, as such they connect at both ends.



>
> [End of technical discussion. Political jeremiad follows. Feel free to ignore.]
>
> With my data consumer hat on, I want to be able to consume the data
> describing these situations without having to guess, because computers
> are very, very bad at guesswork.
>
> With my mapper hat on (the current one is a faded and stained baseball
> cap bearing the logo of a trail crew), I want to produce data that
> will be useful to consumers. I am told that there are other mappers
> who are more 'artists' interested in describing the world around them
> without regard to utility, but I think that most mappers would agree
> with me in wanting the data to be useful to others.

I have a long walking route that has signed access tracks that carry signage that says "Great North Walk".
They are not on the actual Great North Walk, the signs would be better as "To the Great North Walk".
 From time to time they get added to the route which confuses the user.
There needs to be a role for these so they can be recognised for what they are - access tracks.

>
> That's why I'm interested in hearing specificallly from data consumers
> - just as mappers have the actual knowledge of what's in the world,
> consumers have the actual knowledge of how to aggregate it and present
> it to others in a useful fashion.
>
> I'm willing to conform with reasonable mapping guidelines. Where I've
> objected in the past to specific mapping guidelines, it's because of
> one of these sorts of things occurred:
>   - the guideline amounted to 'don't map that!'

And no way to map it... very poor practice!
First tell the mapper how to map it!

>   - the guideline was self-contradictory

Yep. 2 or more editors ...

>   - the guideline provided me no way to distinguish features that my
> rendering needed to distinguish. (That is NOT 'tagging for the
> renderer,' guys!)
>   - the guideline provided me no way to 'rough out' features that I'd
> observed in the field without needing to return for more observations.
> (Or worse, provided no way to map the object without doing research at
> home because information needed for conformance was not observable in
> the field.)
>
> For many 'it's too much work for the mapper' objections, the solution
> is 'make the editor smarter', and the maintainers of the editors have
> been quite good about responding to tagging schemata that actually
> solve a problem and come into common use.
>


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging