OR: OR18/OR99W Error

[michih]

http://travelmapping.net/logs/siteupdate.log

Code: [Select]

Setting up for graphs of highway data.
Full graph has 524578 vertices, 594683 edges.
ERROR: segment name mismatch in HighwayGraphCollapsedEdgeInfo: edge1 named OR99W edge2 named OR18,OR99W
Edge compressed graph has 384637 vertices, 454742 edges.
Is there any combination to the hidden junctions or what's wrong here?

[yakra]

That is due to the HIDDEN_TERMINUS error.

OR18 ends at a hidden point while concurrent with OR99W.
This point is also hidden on OR99W.
We have a hidden vertex with two incident edges -- (edge1 named OR99W edge2 named OR18,OR99W) which HighwayGraphCollapsedEdgeInfo tries to compress into a collapsed edge -- at which point it discovers there's something fishy going on.

[michih]

Ok, that means the error must have exist for a long time. No combination to your latest [edit: site update script] modifications

[yakra]

Indeed it has.

[Jim]

It's been there about as long as the collapsed graph format has been around, which is a few years now.

If OR 18 really ends there, it seems it should be a visible point.

[oscar]

According to OSM and Google Maps, OR 18 peels away from OR 99W at that hidden point, and continues east on the Newberg Dundee Bypass to OR 219. Maybe, besides un-hiding the point on OR 18 and 99W, OR 18 needs to be extended, and OR 219 needs to be updated.

[Bickendan]

Local files have been updated for a while, but I haven't relearned how to use GitHub in conjunction with keeping shared files synced up (eg, the updates file).

[Bickendan]

Essentially, the OR 18 ending on the 99W concurrency on a hidden point was a temporary extension I'd done prior to the Newberg-Dundee Bypass's opening once the OR 18 extension had been formalized. After that, there was an update that I uploaded to Git that didn't work and things got put on hold on my end for a while, then the Bypass opened, and I'm at the point I need a refresher with Git lest I really foul something up.

But the good news is, both OR 18 and 219 are correct on my local side.

[Jim]

A morning site update is underway that includes fixes from @Bickendan for this.

[oscar]

It looks like the latest site update included a correction to OR 219 (rename of OR99W_W, with the old label not kept as an alternate label) that broke my list file. I think I've fixed the problem in the local copy of my list file.

I think something like that happened with the east endpoint for OR 86. I had an extra line in my list file anticipating the change that was just made, but other users might not have been so lucky.

[Bickendan]

I'll need to double check 219, as the only change I've done to it was adding in OR 18 (which apparently threw in a sharp angle error o.0).
As for 86, I think that was a big endpoint switch, eg, ID/OR -> OR/ID with a different crossing into Idaho.

[Bickendan]

OR 219 issue resolved and submitted.