cyclestreets / tflcid-conversion

Conversion of TfL CID attributes to OSM
https://bikedata.cyclestreets.net/tflcid/conversion/
11 stars 3 forks source link

Bicycle dismount (UK 'cyclists dismount' sign) #25

Closed mvl22 closed 1 year ago

mvl22 commented 4 years ago

Richard Fairhurst wrote in: https://lists.openstreetmap.org/pipermail/talk-gb/2019-September/023522.html

A possible but nasty workaround would be e.g. bicycle=dismount_uk representing the UK meaning of a Cyclists dismount sign.

"That is definitely nasty. :) bicycle=advisory_dismount would be more universal."

mvl22 commented 1 year ago

bicycle=advisory_dismount

Personally I like this.

@systemed, can I suggest we have a video call discussion about this - as implementors of two engines, we are probably overdue to have a discussion on this general topic of mutual interest.

systemed commented 1 year ago

On reflection I think this is better as bicycle:advisory=dismount. It chimes with maxspeed:advisory which is quite a popular tag, and degrades gracefully. bicycle=advisory_dismount could conceivably be parsed as "unknown access option so assume no".

mvl22 commented 1 year ago

On reflection I think this is better as bicycle:advisory=dismount. It chimes with maxspeed:advisory which is quite a popular tag

Having now seen the wiki documentation for maxspeed:advisory, I agree this does seem highly analogous and an optimal solution.

The key insight here is that it is not an instruction that sets a legal requirement, and is thus not really a tag in the access family (i.e. bicycle). bicycle=... clearly defines a 'Legal restriction for cyclists'.

Use of the colon makes it a different tag while also keeping the hierarchical implication that it is a second-order matter related to the main access aspect.

bicycle=advisory_dismount could conceivably be parsed as "unknown access option so assume no".

I agree that it would be more likely that a data consumer would parse this as an ignored option. Our own codebase would indeed do that. So yes, this would degrade gracefully.

mvl22 commented 1 year ago

I have now posted to talk-gb on this proposal:

https://lists.openstreetmap.org/pipermail/talk-gb/2022-September/029420.html

mvl22 commented 1 year ago

No objections so far, and only support.

So I propose, unless clear objections start to come in:

Action:

systemed commented 1 year ago

SS_ is from the signage dataset which we don't import (and which I don't think it would be practical to import, given that there's no directionality or other geographic scope beyond the point).

The two places where we have bicycle=dismount are in crossings (if CRS_PEDEST is set) and restricted routes (all routes, which are additionally highway=footway).

Pedestrian crossings really are 'bicycles dismount' so I think we leave the existing tagging there.

Restricted routes include things like the parks, which again really are 'bicycles dismount', so the existing tagging seems correct.

We don't write bicycle=no anywhere.

So I can't see anything we need to do here. Or am I missing something? (always very possible!)

mvl22 commented 1 year ago

Yes, good point.

Anyway this is at least decided on if someone wants to implement the signage dataset in future, and has generated a useful tag consensus more generally.