[Talk-GB] help with reverting changeset (all cycleways in a particular area deleted)

[Andy Townsend]

On 18/04/2022 12:18, Jon Pennycook via Talk-GB wrote:
> I have had to escalate to the DWG - AR_Mapper is really unhappy about 
> the use of the highway=cycleway tag and mapping cycle paths.
>
Hello, Andy from the DWG here.

It's not the first time that there's been a conflict over "too much 
information".  I think the general rule has to be "if you find yourself 
thinking about deleting something that someone has spent a lot of time 
adding, then at the very least you need to discuss it more widely".  
Looking at for example 
https://overpass-api.de/achavi/?changeset=119816211 , it does look like 
there was a deletion like that.  As an example, 
https://www.openstreetmap.org/way/393020953 was deleted, and that was 
clearly part of showing in detail how pedestrians and cyclists were able 
to negotiate the roundabout at 
https://www.openstreetmap.org/way/149533310 , and there wasn't even an 
attempt to add a sidewalk tag to the roundabout itself.

So does that mean that mapping anything as "cycleway=track" (like e.g. 
https://www.openstreetmap.org/way/2987999 ) is "wrong"?  No *1, but most 
roadside cycleways these days (in England at least) do seem to be mapped 
separately.  It's less clear-cut for regular non-cycleable sidewalks *2 
- there's more discussion about that at 
https://community.openstreetmap.org/t/is-there-consensus-on-mapping-pavements-sidewalks-separately-to-roads/1067/10 
(the short answer there is basically "no, there is no censensus").  
However, if someone has mapped separate footways (with all the legal 
links across the road, including whether they include a kerb) then it 
makes no sense to remove that.

One place where https://www.openstreetmap.org/changeset/119816211 wasn't 
over-zealous was around 
https://www.openstreetmap.org/way/582357316/history#map=19/51.42398/-0.73509 
- looking at that in achavi, it did look as if there were rather more 
roads previously mapped than actually existed, but 
https://www.openstreetmap.org/changeset/119874561 seems to have resolved 
that.

One thing I'm not sure we have a consensus yet on is roadside cycle path 
naming.  As an example, 
https://map.atownsend.org.uk/maps/osm-deep-history/#/way/393273071 is a 
roadside cycle path along a road called "Harvest Ride". Should it have 
no name (because it itself doesn't have a name, although the adjacent 
road does), the name of the road, or something based on the road like 
"Harvest Ride cycle path"?

Hopefully we can get all sides of the discussion here.  I know that the 
"other" mapper in this example was active on Discord and I mentioned 
that this discussion was happening; here's hoping that they'll manage to 
subscribe and  post to the list as well.

Best Regards,

Andy

*1 although a wider look at usage e.g. https://overpass-turbo.eu/s/1hRG 
suggests some "cycleway=track"tags are misused, or at least don't look 
like https://wiki.openstreetmap.org/wiki/Tag%3Acycleway%3Dtrack .

*2 leaving aside the all-too-common case of a local authority pretending 
to provide cycle infrastructure by declaring existing narrow sidewalks 
as "cycleways", saving cyclists from marauding cars but leaving 
pedestrians at the mercy of cyclists.