2023-02-21 | 2023-02-22 | 2023-02-23

2023-02-22

Liverpool

Last night, I ended in Liverpool by requesting all AS30's inside my expanded radius. How did they get on?

It's still processing! Wow. That's more errors than I'd like, though. Let's investigate.

1. Error investigation
   • Count the errors:

     lhc@S23:~$ clear; find /var/www/html/Oasis/tmp/A*Failure/ -type f  | wc -l
     
     42
     

     • That means there's actually 21 errors, because each one has a .json file and a .stderr file. 21 is the amount of failed AS30's I can see on the map, so that's reassuring, in a way.
   • Investigate the errors:

     lhc@S23:~$ clear; find /var/www/html/Oasis/tmp/A*Failure/ -type f  | sort
     

     • This involves a lot of cat 'ing .stderr files. It's not particularly exciting, so I'll spare the nitty-gritty. In short, all but one of the AS30's failed because of TopologyExceptions - this is usually lines that are too close together, or ways that cross without a node at the junction, etc. One particular AS30 failed, citing this reason:

       OasisOSM.py                002121 ERROR: Calculated height is not a number
       OasisOSM.py                002121 ERROR: Calculated height is not a number
       OasisOSM.py                015906 ERROR: Node[25573] is inside 2 WATER polygons (0 or 1 expected)
       AS30Create.sh              000276 ERROR: File /var/www/html/Oasis/Data001/M30/53_15_00_N_002_45_00_W/53_26_15_N_002_57_45_W/53_26_15_N_002_57_45_W_F056_Aerodrome.json does not exist
       

     • That's interesting.
2. Error Investigation - Deep Dive
   • To investigate the above issue, I'm going to run OSM-MeshJSON on the command line (without --doit). THis allows me to see the exact output I get for this AS30, without recreating it and clogging up our OSR's.

     lhc@S23:~$ OSM-MeshJSON --as30refs 53_26_15_N_002_58_15_W
     

   • Straight away, I have a new multi-classification to address:

     <3>015066 ERROR: Element multi-classified as ['RAILWAYAREA', 'SCRUB'] (type=way id=180893494)
     

     • Fixed that - I chose to keep the scrub.
   • I've also been doing some OSM work - there's a few bad practices:
     • Lines crossing way boundaries without a junction node;
     • Ways crossing into each other!
     • Ways with ways inside them, that aren't relations...
     • The list goes on.
   • In some instances, I've fixed the issues, in others, I've reduced the AS30's BufferArcSeconds. Hopefully tomorrow I'll be in!

Dublin

Previously... Oasis.2023-01-31#Regression_Testing_45_Dublin

I want to see how Dublin looks with 'flat water' instead of the OSM height override, so I need to remove the OSM array from all the affected AS30s. So I don't forget, this what the JSON looks like:

   "OSMHeights": [
      {
         "Height": 0.0,
         "areashrink": -0.15,
         "id": 47864604,
         "type": "way"
      }
   ],