This topic concerns Helen Street in the City of Parramatta NSW. I have run the full length of this street (purple line right along it), it only has two nodes, one at each end, and both are green. However, it is still showing in my Incomplete list rather than the Complete list.
I would be interested to know if there is some issue here. My profile is at Mel Cox - CityStrides.
Regards
Mel
Edit: I think I have found the reason - there are two Helen Streets in the City of Parramatta (and I haven’t run the other one)!
Ugh, yeah, this is a bummer. It’s talked about elsewhere in the forum, but I don’t have time to search right now - basically OSM doesn’t know what a “street” is in the way that we do. It only knows about “ways” which are sections of a street - might be the whole thing, might just be a curve, etc.
In order to cobble together the idea of a street, I have to match up the individual ways somehow. I’m using the name field to do this, which causes this issue of multiple same-named streets in one city being packed together as one street record.
There are some approaches that might help avoid this, but nothing is perfect. Like, collecting all the ways where the nodes are “close” to each other - for various definitions of “close” - will sometimes mistakenly separate streets … or the definition of close will be too near, and it’ll still pack different streets together.
Here in Liverpool the majority of duplicate street names (same name, different location) are separated by some distance. I tend to think that moving to a system that creates separate street where they have the same name but are separated by distance would create a more accurate count than the current system of having separate streets counted as one.
The problem is that the data from OSM doesn’t support keeping separate streets with the same name separate. The data from OSM consists of nodes with street names. How far should two nodes be before CS considers them separate streets? And since the database is based on street names and cities, how do your keep the two entries separate.
James said the same thing about OSM data and how it handles street names. He did suggest he could force CS to treat roads with the same name but separated by a distance (to be decided) as separate roads.
Until one long straight road comes along, who’s two nodes are further apart than the “not the same road/way” threshold.
Personally, I don’t think it’s a huge deal. I assume we are going to run all the streets, whether they have the same name or not. In fact I’ve done it several times on “disconnected ways of the same name.”
Bigger fish for James to fry this days.
Collect the nodes, complete the streets… Even if they are in different parts of the city.
OSM groups nodes into ways, streets can be made up of any number of ways that have the same name. Knowing whether the ways form one street or several that just happen to share the same name is not trivial. Grouping by proximity would probably work for most, but there are cities like Arlington, VA, that have streets split into disconnected chunks where this approach would create extra streets…
I’ve been having a similar issue, street is at 100% but not showing as complete. No duplicate street names in this town Stratham, NH. Chris Ager - CityStrides
Barnes Drive is an example, there have been 20-30 other streets doing the same thing over a week’s worth of runs.
When was the city last updated? There was a bug that would correct the nodes but would not update the number of nodes. This means that you could turn all the nodes green, but since there were only (for example) ten nodes but the database claims there were twenty, you couldn’t finish the street.
This bug has been addressed (I think), but if the city hasn’t been updated recently then the fix hasn’t been applied to your city.
I have streets where all the nodes are green or over 90% are in cities that were updated in the past couple of days, but the street is still marked as incomplete
Apparently there’s still some tweaking that needs to be done to the update procedure. (I noticed recently that the street count is inconsistent on my city.)
I have no idea how long things take, but the problem reported by @crewchiefkc135 (Chris) on 12/20 is no longer there. All streets in Stratham show complete. And by the way Chris, you have an impressive list of 100% cities!
I’m still waiting for an update for Keller, TX (last was 2020-10-20). I don’t the know the order, but several cities near me have updated recently.
PS. Keller has 10 streets with the 0-node problem. I have run them all, so hopefully I will be 100% in Keller with the next OSM update. Oh, and sorry for dragging this off subject. I got sidetracked and started typing. Any post on updates catches my eye.
I have an example here, Småbåtsvägen in Lidingö,
It had 16 nodes, but half of it is behind a closed gate. So in december I edited OSM to indicate private. Now when Lidingö was updated in CS on Jan 28 this should have been reflected? But now I see in the map that it has seven nodes, all green for me. But it still seems to expect 16 nodes, so listed as incomplete
I have the same after a city update on several streets that were set to private. The total node count for each street seems not to be adjusted to the new runnable nodes
Thanks for the link - this is a perfect example of both a street with incorrect nodes count & a city with incorrect streets count … both issues will affect progress calculations.
I’ve just ran a city update to see how the updated code does… It correct both counts.
The progress calculation will take a little more time - that goes through a different, slower, queue. Let me know if you see your percentages change for better/worse today.
Ugh! I had some more post-deletion cleanup to do in my code - I’m pretty sure I just found the last couple places where I’m deleting data & not updating counts.
The woes of scaling massive amounts of work by optimizing code instead of spending more money on servers/databases.
The woes of scaling massive amounts of work by optimizing code instead of spending more money on servers/databases.