[OHM] evolving road classifications & OHM

Richard Welty rwelty at averillpark.net
Mon Jan 6 23:16:17 UTC 2020


On 1/6/20 5:10 PM, Rob H Warren wrote:
> Richard,
> 
> You know the OSM data model does not deal well with change and has poor (geometry<>feature<>label) separation. My suggestion is that since the roadway has not changed significantly, have the geometry encoded as a series of nodes that get allocated to different temporal ways for the different "features" of that route over time. 
> 
> It is not a great solution but it works and is compatible with the renderers. The properties of roadways (width, etc...) are encoded at the way levels in OSM which lets you reuse the OSM highway tagging system.

based on my experiences in the past with JOSM, i think such an approach
would be pretty error prone and difficult. over on Slack, Leon Karcher
suggested tagging the single way with its most appropriate recent tags
and using relations for the older one.

> Additionally, you could use relations to link the different ways in a temporal/sequential fashion (eg: nextIncarnation). I've experimented with a similar solution for historical LOD and wikipedia does the same for elements that have multiple incarnations. I don't think that this would break anything in the ecosystem since unknown relations get ignored.

more or less what Leon suggested, and what i was seriously considering
as an approach.

i am not worrying over much about the rendering on the main website, as
in its current form it has somewhat limited value, but about tagging so
that overpass queries can be formulated in useful ways to feed into what
ever kind of apps and widgets we develop for use of this data.

which leads me to...

way way back in 2015, this section of the OHM wiki page got set up:


https://wiki.openstreetmap.org/wiki/Open_Historical_Map#Short-term_plans

the link to OHM/Tags was never filled in in the 4+ years since. i think
it's high time we started describing how people are tagging to reflect
historic features and dating. i certainly am working through a lot with
my stuff, and the method Leon is using could certainly go in as well, as
examples of experiments as this is largly experimental at this time.

richard
-- 
rwelty at averillpark.net
 Averill Park Networking - GIS & IT Consulting
 OpenStreetMap - PostgreSQL - Linux
 Java - Web Applications - Search



More information about the Historic mailing list