-
Notifications
You must be signed in to change notification settings - Fork 659
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Ferry Reclassification - Can reclassify nearly all edges on small islands #4358
Comments
if we dont reclassify them the island becomes a "destination (or origin) only" island right? this would essentially make rural ferry ports unusable on longer routes, which might be ok but it would be interesting to see on how many ferries the reclassify algorithm would give up. we need to log these so we can at least quanitfy the problem and maybe hone in on a magical value that works for most ports. maybe its 32, maybe its 42, maybe 111? also maybe it might be better to base it on distance rather than on number of nodes. after we fix the other issue were we only upclass the shortest path, if we find that the expansion has made it 10 miles (or whatever number) away from the ferry or something maybe its best not to upclass anything there. |
The answer is 42. Always. |
The main reason why the referenced condition in there like that, seems to be that some ferry connections are high class but only connect to low classes further inland, right? (FWIW, the Tsawwassen terminal is only trunk roads these days) Couldn't we do smth like we do for
that wouldn't hold of course, but honestly, that seems pretty messed up. We'd need to protect against ferry terminals like Tsawwassen where the ferry directly connects to the full trunk network, e.g. if we saw more than 20-30 primary edges in a row, we'd break and assume we don't need to upclass. It's not accurate either in all situations but I think it might lead to less problems. Then we wouldn't have to care about any random limit of found primaries or even number of iterations. Any limit would (and currently does) very likely also cause non-shortest paths to be upclassed if the 33rd primary road is randomly somewhere pretty far away (though a distance instead of node limit would alleviate some of that). And if there's no primary to be found, the adj list is run empty and we don't upclass anything (I agree, that's a gross oversight that we still do upclass currently). |
Hm, just seeing that we don't even try to reclassify if the connection edge is a primary.. I wonder what Tsawwassen looked like before.. Still, the thought is still similar: if we expand on more than 20-30 consecutive primary (or higher) edges on the same expansion branch, we assume it's fine, break and reclassify? |
I found a case in NY (other cases may be likely) on Fishers Island, NY (south of New London, Connecticut) where all edges are reclassified when reclassifying edges connected to ferries. This island has no ways/edges that meet the target classification (primary) - so the algorithm continues to expand from the ferry and essentially reclassify nearly all edges. The algorithm continues until the adjacency queue is empty or:
the 2nd half of this condition is never met on Fisher Island.
I think we should:
The text was updated successfully, but these errors were encountered: