[talk-ch] Changes to HikingNetwork wiki page

Sarah Hoffmann lonvia at denofr.de
Sat Sep 4 00:19:25 CEST 2021


On Fri, Sep 03, 2021 at 11:12:14PM +0200, Yves wrote:
> Why not just osmc:symbols=yellow_diamond on the ways?
> *=designated would be more for acces rights.

Because using anything but relations creates an exception that data users
have to know about and handle.

Sarah

> Yves 
> 
> Le 3 septembre 2021 21:09:24 GMT+02:00, Raphael <dafadllyn at gmail.com> a écrit :
> >Hi Matthijs, hi everyone
> >
> >It seems that a consensus hasn't been reached yet - at least not on this list.
> >
> >The Swiss hiking network is a bit peculiar. It's not a typical node
> >network as the intersections (or nodes) aren't numbered and very often
> >are unnamed. On the other hand the hiking trails (not including the
> >routes from SchweizMobil) aren't typical routes as they are unnamed,
> >unnumbered, have countless possible starts and ends and often have
> >multiple variants from one named place to the next.
> >
> >In my opinion, the simplest way to map the Swiss hiking trails is by
> >adding a tag to the corresponding paths (maybe trail=hiking or
> >hiking=designated) and using route=hiking routes only for the "real"
> >SchweizMobil routes.
> >
> >Best regards
> >
> >Raphael (dafadllyn)
> >
> >On Mon, 23 Aug 2021 at 20:21, Matthijs Kooijman <matthijs at stdin.nl> wrote:
> >>
> >> Hi all,
> >>
> >> I'm new to this list, so let me start with a small introduction: I'm a Dutch
> >> mapper, but I visit Switzerland every now and then (my brother lives in Rieden
> >> SG) and I like to do some hiking and mapping here. In the Netherlands, I've
> >> done some mapping on the cycle node networks, and in Switzerland I like to work
> >> on the wanderwegen network (though I have not mapped much, so treat my input
> >> accordingly).
> >>
> >> I've been involved on the wiki page discussion section before, and from
> >> the wiki page history found a link to this thread, which I have read
> >> with much interest. Hopefully I can add something to the discussion in
> >> this thread.
> >>
> >> From reading this discussion, it seems that there are still some matters
> >> where no consensus has been reached (in particular the matter of
> >> overlapping or non-overlapping routes). But it also seems that there has
> >> been some discussion about this off-list. Maybe some consensus has been
> >> reached elsewhere?
> >>
> >>
> >> In case you have not seen it yet: there has been some effort (initiated by
> >> Peter Elderson AFAIU) to document current practices and rationale about node
> >> networks. This is mostly based on the numbered cycle/walking networks in NL/BE,
> >> but explicit effort has been made to also include named networks like Germany
> >> and Switzerland uses. AFAIU there has already been some experiments involving
> >> parts of the german network.
> >>
> >> The current state of this documentation can be found here:
> >>
> >>         https://wiki.openstreetmap.org/wiki/Node_Networks
> >>
> >> If this has not been done already, I would suggest considering to make the
> >> Swiss tagging scheme either conform to that page, or modify/generalize that
> >> page to also apply to the Swiss tagging.
> >>
> >> I just read through the page, and here's some notable observations from the
> >> current Swiss practice:
> >>  - The primary goal of the model specified is to facilitate routing and
> >>    rendering of node networks. For example, the knooppuntnet.nl website allows
> >>    entering a starting and ending node, and calculates a route between them
> >>    using only routes part of the network, and produces a list of node numbers
> >>    to follow (support for names instead of numbers is nearly complete, I
> >>    believe). For the Swiss case, this could work the same, but produce a list
> >>    of named nodes/guideposts the route passes (so you could walk the route
> >>    without bringing an actual map, just follow signs to each of the names in
> >>    turn).
> >>  - That page specifies to tag the actual junctions (as part of the ways they
> >>    connect), instead of (or in addition to) the physical guideposts (besides
> >>    the way they connect), which is a significant difference from the current
> >>    Swiss practice. I believe the goal is to simplify routing, since you can
> >>    then know which named junctions you pass exactly, without needing to find
> >>    guideposts based on nearness. It can also help with network consistency
> >>    checks (especially when combined with the expected_STn_route_relations
> >>    attribute).
> >>  - That page also talks about junctions without a number or name, and specifies
> >>    to create multiple partially overlapping routes in that case (and tagging
> >>    the unnamed nodes with `xxn_ref=*`).
> >>    Note that for the Dutch case, unnumbered junctions are very rare and almost
> >>    always in the near vicinity of a numbered node, so in the Swiss case in
> >>    regions that have more unnamed junctions, overlapping routes might have
> >>    additional downsides (such as a potential explosion of possible routes when
> >>    multiple unnamed guideposts exist between named routes, and the extra
> >>    difficulty mapping partial routes).
> >>  - That page specifies to use `name=from-to` for named node networks (numbered
> >>    networks use `ref=from-to`), while discussion on this list was moving
> >>    towards not using the name anymore.  However, it might be that that page
> >>    specifies this to match current practice and could be changed without
> >>    problems.
> >>  - That page specifies a "network" relation (not strictly required, though),
> >>    that collects all the routes and nodes belonging to a specific node network,
> >>    to allow semantically grouping routes and specifying e.g. the operator once
> >>    for the entire network. In the Swiss case, I think this could amount to one
> >>    network relation for each Canton.
> >>  - That page specifies a `network:type=node_network` attribute for nodes,
> >>    routes and networks, to help distinguish them from regular routes.
> >>
> >> Some earlier discussion about supporting named node networks has been done here:
> >>         https://github.com/vmarc/knooppuntnet/issues/102
> >>         https://wiki.openstreetmap.org/wiki/Proposed_features/Named_nodes_in_node_networks
> >>
> >>
> >> Then, reading the wiki page in its current state, it seems to better reflect
> >> current and intended practice, which is good. However, I think there are still
> >> some things that could be clarified. In particular:
> >>  - How to tag unnamed guideposts? Common practice seems to be to just omit
> >>    name, maybe that should be explicit on the wiki?
> >>  - How to handle numbered guideposts? For example in Graubünden (I walked in
> >>    the area around Trin), I've seen guideposts (named and unnamed) that have a
> >>    three-digit number that were already mapped with the number in the `ref`
> >>    attribute, maybe that would be good to add to the wiki?  For an example
> >>    (with photo) see: https://www.openstreetmap.org/node/6100531211 Similarly, I
> >>    understand that newer guideposts have a 6-digit unique number.
> >>  - Would it make sense to put these guideposts numbers in the relations?
> >>    Especially when using the "non-overlapping routes between junctions"
> >>    approach, putting these numbers in the from and to attributes could help
> >>    diagnose issues. OTOH, they are mostly meaningless to users, in the sense
> >>    that the signing does not point *to* these numbers, you can only see them
> >>    when you are at a particular guidepost.
> >>  - How to tag from/to with unnamed guideposts? Should we invent names?
> >>    Just omit the tag? I think no full consensus might have been reached
> >>    here?
> >>  - Maybe make explicit that name= was used with from-to before, since a
> >>    lot of existing routes still have that. If it is explicit that this is no
> >>    longer recommended, that will be less confusing for new mappers.
> >>  - Is it ok to copy routes from the cantonal GIS systems? If not, or
> >>    only from some, maybe make this explicit?
> >>
> >>
> >> Gr.
> >>
> >> Matthijs
> >> _______________________________________________
> >> talk-ch mailing list
> >> talk-ch at openstreetmap.ch
> >> http://lists.openstreetmap.ch/mailman/listinfo/talk-ch
> >_______________________________________________
> >talk-ch mailing list
> >talk-ch at openstreetmap.ch
> >http://lists.openstreetmap.ch/mailman/listinfo/talk-ch

> _______________________________________________
> talk-ch mailing list
> talk-ch at openstreetmap.ch
> http://lists.openstreetmap.ch/mailman/listinfo/talk-ch



More information about the talk-ch mailing list