[Tagging] The direction=* tag

Tod Fitch tod at fitchdesign.com
Fri Mar 17 15:19:54 UTC 2017


Thank you for the insight. I don’t know that it is off topic though as any tagging scheme should be both achievable by a typical mapper and useful to any anticipated data consumers. Something that is easy for taggers but not useful to data consumers is just noise. Something that is useful for a data consumer but hard for a mapper is, well, a hard problem.

In this case, it seems that adding a “direction = forward | backward” tag to a node also tagged with “highway=stop | give way” is just noise.

> On Mar 17, 2017, at 7:38 AM, Daniel Hofmann <hofmann at mapbox.com> wrote:
> 
> On the routing engine side these tags will be translated to turn penalties for specific turns, so the matching equivalent would probably be from,via,to relations. Which is not a feasible modelling in OSM realistically speaking. Simple stop signs could go on ways, but fail to capture more complex situations.
> 
> The problem with traffic lights specifically is their interconnectivity: you don't want to add turn penalties multiple times if traffic lights are connected. But you can't tell. This is a bit off-topic, feel free to cross-post to the osrm-talk or routing mailing list, or open an issue in osrm-backend.
> 
> On Fri, Mar 17, 2017 at 3:29 PM, Tod Fitch <tod at fitchdesign.com <mailto:tod at fitchdesign.com>> wrote:
> From a data consumer point of view, what type of tagging would be reasonable to indicate the, for lack of a better work, direction of travel a stop, give way or traffic light has effect?
> 
>> On Mar 17, 2017, at 3:54 AM, Daniel Hofmann <hofmann at mapbox.com <mailto:hofmann at mapbox.com>> wrote:
>> 
>> Jumping in here to give a perspective from a routing engine (OSRM, https://github.com/Project-OSRM/osrm-backend#open-source-routing-machine <https://github.com/Project-OSRM/osrm-backend#open-source-routing-machine>). We do not handle direction tags on nodes which indicate a property for a way or a turn at an intersection. The example with stop signs and give yield signs is spot on. Even worse is the assumption that routing engines can just infer the direction by checking the distance to the nearest intersection. This is in conflict with how parsing and creating a graph works.
>> 
>> There is a similar problem with exit_to node tags, indicating the exit way destination - you can read about it here
>> 
>> - https://www.openstreetmap.org/user/daniel-j-h/diary/40555 <https://www.openstreetmap.org/user/daniel-j-h/diary/40555> (en)
>> - https://www.openstreetmap.org/user/daniel-j-h/diary/40554 <https://www.openstreetmap.org/user/daniel-j-h/diary/40554> (de)
>> 
>> Cheers,
>> Daniel J H
>> 
>> 
>> On Fri, Mar 17, 2017 at 11:25 AM, Martin Koppenhoefer <dieterdreist at gmail.com <mailto:dieterdreist at gmail.com>> wrote:
>> 
>> 2017-03-16 5:13 GMT+01:00 Tod Fitch <tod at fitchdesign.com <mailto:tod at fitchdesign.com>>:
>> The “direction” tag [1] has different uses that seem disjoint to me.
>> To specify the orientation (compass point or degrees from north) of an object (adit or cave entrance, etc.). 
>> 
>> "orientation" would have been a better descriptor IMHO, but the crowd uses this tag differently (see taginfo, also subtags like roof:orientation, ...). Direction is working for me nonetheless.
>> 
>> 
>> 2. To specify direction (clockwise/counterclockwise) around a roundabout (not sure why this is needed as it should be apparent from local laws or specified with a “oneway=yes”).
>> 
>> 
>> agree with you
>> 
>> 
>> 3. To indicate the direction (forward/backward) a stop or yield (give way) sign has effect along a way.
>> 
>> 
>> broken. From time to time people are coming up with features to tag on nodes that require (or seem to require) the information of a direction. Taking the direction of a different object (e.g. here a way) doesn't seem a healthy way to represent this. Ways might get split, might get reversed, nodes might be (or become) part of several ways, etc. Either use a cardinal direction or a short way stub or a relation, etc., but not "forward" or "backward" tag values on a node, it simply doesn't make sense. Tags should refer to the object they are tagged on.
>> 
>> 
>>  
>> Oddly, that third use seems only for stop and yield signs but not for traffic signals where a “traffic_signals:direction=forward | backward” tag is to be used. However that seems to be the most used form [2]. Apparently some have figured that if we have “traffic_signals:direction” there should be “stop:direction” [3] and “give_way:direction” [4] tags.
>> 
>> 
>> similarly broken
>> 
>>  
>> I would keep the variant 1 and discourage 2 and 3.
>> 
>> Cheers,
>> Martin
>> 
>> 
>> 
>> _______________________________________________
>> Tagging mailing list
>> Tagging at openstreetmap.org <mailto:Tagging at openstreetmap.org>
>> https://lists.openstreetmap.org/listinfo/tagging <https://lists.openstreetmap.org/listinfo/tagging>
>> 
>> 
>> _______________________________________________
>> Tagging mailing list
>> Tagging at openstreetmap.org <mailto:Tagging at openstreetmap.org>
>> https://lists.openstreetmap.org/listinfo/tagging <https://lists.openstreetmap.org/listinfo/tagging>
> 
> 
> _______________________________________________
> Tagging mailing list
> Tagging at openstreetmap.org <mailto:Tagging at openstreetmap.org>
> https://lists.openstreetmap.org/listinfo/tagging <https://lists.openstreetmap.org/listinfo/tagging>
> 
> 
> _______________________________________________
> Tagging mailing list
> Tagging at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20170317/42bf7394/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 1861 bytes
Desc: not available
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20170317/42bf7394/attachment.bin>


More information about the Tagging mailing list