Closed matthiasfeist closed 2 years ago
Yes, we should probably do that. I've also noted that oneway with lanes=1 may be desirable on large roads, as it seems to be always set in current OSM data, so maybe always keep lanes.
I want to think a bit more about this issue, haven't merged it yet. It will lead to that we add lots of lanes=2/lanes=1 where there is none currently. I'm not a big fan of adding redundant tags. I'm thinking about that this should only be used on larger roads... but I'm not sure. I did not actually know about the existence StreetComplete until recently.
Makes sense.
StreetComplete is a really great tool for casually improving OSM on the go. I use it a lot :)
I was just checking the wiki on this issue and I saw that it's stated there that:
Although it is advisable to always add the lanes tag to roads with lanes, as usual data will be sometimes missing
I would therefore say that it's actually OK and seems to be wanted to have the lane count explicitly in the tagging all the time.
I guess "road with lanes" means roads that have line markings to actually show the lanes, but not unpaved roads.
NVDB has lane info but not for all roads. I think it's only the government maintained roads if I remember correctly. It has also lanes on unpaved roads which then is directly related to width, so it's 2 if the road is wide enough, otherwise 1. Track, service, unclassified and residential typically has no lane info in NVDB though if my memory is correct.
One idea would be to keep NVDB lane info as is, but not make up lane info if it doesn't exist.
if we can figure out that there are no markings, we could tag it accordingly: https://wiki.openstreetmap.org/wiki/Key:lanes#No_lane_markings
but I agree that maybe it's best to keep the info from NVDB as it is...
I have committed a patch which keeps lanes=1 and lanes=2 when set by NVDB.
cool.
Hi. I have a suggestion to keep the lanes tag even if lanes=2 and not oneway. Code here: https://github.com/atorger/nvdb2osm/blob/4538761f0f6dcf4fa958697b5baf2f3c6798126b/process_and_resolve.py#L1751
My reasoning because I made 2 observations. Both are not because of a right or wrong way of tagging but they mostly evolve around the UX of editors:
What do you think?