Hi,

Sorry for going back and forth on this one. Part of the difficulty is communicating with everybody editing in Seattle and building consensus. I started some threads in the OSMUS Slack about it, https://osmus.slack.com/archives/C1FKE1NCA/p1624292886016000 and https://osmus.slack.com/archives/C1FKE1NCA/p1627404161080800, and got some agreement that emphasizing bicycles over pedestrians here was strange. Unfortunately I think you have to sign up for Slack just to view the links, and I realize that's not really a canonical forum -- it's just a convenient one. Do you frequent any mailing lists or other places where it's easier to discuss stuff like this before making a change?

As a data consumer, I'm just trying to find some way of distinguishing a highway=cycleway like this inner path from the cycletrack on E Green Lake Dr N and other shared-use spaces like the Burke Gilman. The inner trail in practice requires much lower speeds, since the segregation between foot and cycle traffic is not followed in practice, from my experiences there. From the tagging highway=cycleway + foot=designated + segregated=yes, this trail sounds more friendly to faster bike traffic than the Burke (osm.org/way/163631458) with segregated=no, but I would disagree from my experiences there.

Some of the tension may be that the infrastructure doesn't match the use in practice, and some kind of historic data about volume and speed of bicycle traffic along each path might be more helpful for use in routing. But that gets out of the realm of OSM. I'll keep thinking about ways to distinguish these 3 cases (the inner trail, the cycletrack, and the Burke) with OSM tags. If you'd prefer to switch it to highway=cycleway in the meantime, I'd be happy to make that change.

Thanks,
-Dustin

`parking:lane:right = parallel` is a way to indicate the rightmost lane is used for parking

Using separate ways to tag more detail about complex crossings makes sense to me. I deleted the duplicate tags on the road in osm.org/changeset/107444042. I'll keep iterating on my end to snap the separate path to the road and create a simple representation for my purposes. I'm also going to be experimenting with tagging the width and type of buffer around cycleways soon; I'll CC you on those changesets. Thanks for your mapping!

Hi,

I tagged the cycletrack on the road: osm.org/way/6410233. So now the same infrastructure is represented twice. We need to decide which to use. Per osm.wiki/Key:cycleway#Cycle_tracks, a separate way or an attribute of the road are both valid methods -- the separation here is just flex posts and striping. Do you have reason to prefer using a separate way? For the software I work on, attributes of the road are much easier to reason about, so I'm going to have to heuristically snap a separate way on my end.

This is a common problem, by the way -- another example of bike paths marked twice is osm.org/way/948577962 and osm.org/way/621426647. It's quite a headache to fix. I know the default carto renderer doesn't show cycle infra; that's why I've built http://abstreet.s3-website.us-east-2.amazonaws.com/0.2.52/osm_viewer.html?system/us/seattle/maps/downtown.bin&--cam=20.34/47.61504/-122.33544 to help visualize things like this.

I'd love for all the mappers around Seattle to settle on a consistent style for tagging bike infra, so I can stop trying to handle both cases in my software.

Thanks,
-Dustin

Thanks! I tagged many of the dual carriageways I spotted in OSM as split one-ways, but haven't verified those sections actually do have a physical median.

Sorry for the slow response and thanks for this explanation! I won't revisit https://github.com/a-b-street/abstreet/issues/330 (to make sure things like the Burke are modelled reasonably in A/B Street) anytime soon, but I'll revisit your notes here when I do.

Hi,

I'm consuming this data for routing and simulation in abstreet.org. Currently, I'm interpreting highway=footway, bicycle=yes to mean that bikes can use the way, per the "bicyle=yes" note in osm.wiki/Tag:highway=footway. I'm curious what you mean in your changeset comment by the Burke routing being normal again. How do I know this footway is usable by bikes? By the absence of bicycle=no? By segregated=no?

Thanks!
-Dustin

You're correct, I misread the imagery. That's an odd lane placement, but it exists. Should be fixed now. Thanks for catching my mistake!

Whoops, good catch! Trying it out again, ID doesn't warn about this. It's probably time for me to learn josm and opt into more sanity checks.

That should work! Thanks so much for the super quick fix.

Hi,

I noticed osm.org/way/797198032 and osm.org/way/387912052 have duplicate geometry. This is causing some issues in an OSM tool I'm trying to apply to the area, and I'm pretty sure it's an error. Should the construction one be deleted?

Thanks,
-Dustin

Hi,

Before, there was a small way osm.org/way/722013503 whose center-line was a strict sub-set of the larger way osm.org/way/722013169. The smaller one was seemingly created to mark parking:lane:both = no_stopping along that small stretch. If I understand the OSM data model, this wasn't a valid representation, because one stretch of road was represented by two different ways.

So to fix this, I
1) deleted the smaller way
2) split the larger way at the point where the smaller way started
3) marked the smaller way (now osm.org/way/722013169) with parking:lane:both = no_stopping

I could also sketch a quick diagram if that'd be more clear.

Thanks!
-Dustin