Board index FlightGear Development AI Traffic

EDDF ground net  Topic is solved

Intelligent, computer controlled vehicles that drive/fly over the planet!

Re: EDDF ground net

Postby wkitty42 » Thu Apr 12, 2018 1:30 pm

there was 13 of'em stuck... i might have given the first boot to the wrong one... if i can figure out what order they're being controlled in, and give the boot to the one at that head of the line, they might all start going on their own again... the ones at the far end near 18/36 look to maybe have been waiting on others behind them taking a slightly different path to catch up... i don't know and now that i've probably broken them out of order, i'll probably have to stay on them until they depart...

when this all flushes out and all the departing ones are gone, that'll leave just one, EK046, all by himself at EDDF... and just as i was writing that, he requested to startup so he's going to be leaving before too long, too :lol:
"You get more air close to the ground," said Angalo. "I read that in a book. You get lots of air low down, and not much when you go up."
"Why not?" said Gurder.
"Dunno. It's frightened of heights, I guess."
User avatar
wkitty42
 
Posts: 5634
Joined: Fri Feb 20, 2015 3:46 pm
Location: central NC, USA
Callsign: wk42
Version: git next
OS: Kubuntu 14.04.5

Re: EDDF ground net

Postby gooneybird » Thu Apr 12, 2018 2:13 pm

There seems to be no pattern to this blocking. I've been using the Channel Islands for testing while working on the groundnets and sat at Guernsey the last week or so and I've quite often had the first aircraft taxi to the end of the runway and just sit there until I quit and restart FG. On the restart all 12 of the test traffic (1 departing every 5 min.) will usually happily depart on time with no hold ups.

The same goes for the pushback routes, one minute they work perfectly and the next there's aircraft wandering all over the scenery :shock:
gooneybird
 
Posts: 2971
Joined: Sat May 31, 2008 1:57 pm

Re: EDDF ground net

Postby wkitty42 » Thu Apr 12, 2018 2:36 pm

errrrggg... looks like there's still a problem... i got the exceptions and suspension again... i didn't look for it until just now... still not sure if it is a lack of parking spaces or if it is a code problem... it looks like existing flights are still working until they fly out of range (200nm radius) and their objects are removed... while typing this, i heard a flight at Schipol doing its thing but EDDF is empty now except for one craft i saw fly in, land and taxi to his parking spot...

Code: Select all
Traffic manager: Creating AIModel from:EDDF-LIMF.xml
Could not find parking for a 733 of flight type gate of airline     DLH at airport     EDDF
caught exception processing subsystem:traffic-manager
message:Bad arguments to findShortestRoute

Traffic manager: Creating AIModel from:EDDF-UUDD.xml
Could not find parking for a 320 of flight type gate of airline     DLH at airport     EDDF
caught exception processing subsystem:traffic-manager
message:Bad arguments to findShortestRoute

Traffic manager: Creating AIModel from:EDDF-EKBI.xml
Could not find parking for a 733 of flight type gate of airline     DLH at airport     EDDF
caught exception processing subsystem:traffic-manager
message:Bad arguments to findShortestRoute

Traffic manager: Creating AIModel from:EDDF-KIAD.xml
Could not find parking for a 74H of flight type gate of airline     DLH at airport     EDDF
caught exception processing subsystem:traffic-manager
message:Bad arguments to findShortestRoute

Traffic manager: Creating AIModel from:EDDF-LSZH.xml
Could not find parking for a 320 of flight type gate of airline     DLH at airport     EDDF
caught exception processing subsystem:traffic-manager
message:Bad arguments to findShortestRoute
(exceptionCount=5, suspending)


BUT

i've gone digging more and i find a parking.xml file over in %FGDATA%/AI/Airports/EDDF of all places... that one has 118 parking places defined along with more than just parking... it looks similar to the EDDF.groundnet.xml i just downloaded above... a diff (after dos2unix) shows there are quite a few lines that are the same but a lot that are not... i'm moving this parking.xml file out of the way and trying again with just the EDDF.ground.net from above...

cold it be that the parking.xml file i found is the old EDDF.groundnet.xml?? with all those parking spaces, we might not have lost what we thought we lost ;)

testing resumes with the EDDF.groundnet.xml from gooneybird...
"You get more air close to the ground," said Angalo. "I read that in a book. You get lots of air low down, and not much when you go up."
"Why not?" said Gurder.
"Dunno. It's frightened of heights, I guess."
User avatar
wkitty42
 
Posts: 5634
Joined: Fri Feb 20, 2015 3:46 pm
Location: central NC, USA
Callsign: wk42
Version: git next
OS: Kubuntu 14.04.5

Re: EDDF ground net

Postby wkitty42 » Thu Apr 12, 2018 2:40 pm

gooneybird wrote in Thu Apr 12, 2018 2:13 pm:There seems to be no pattern to this blocking. I've been using the Channel Islands for testing while working on the groundnets and sat at Guernsey the last week or so and I've quite often had the first aircraft taxi to the end of the runway and just sit there until I quit and restart FG. On the restart all 12 of the test traffic (1 departing every 5 min.) will usually happily depart on time with no hold ups.

do you mean when they get stuck and need a boot to the backside? yeah, i think that's something in the code... i don't know how it is keeping up with the order of the craft it is controlling... i haven't gotten that far in studying the code, yet...

gooneybird wrote in Thu Apr 12, 2018 2:13 pm:The same goes for the pushback routes, one minute they work perfectly and the next there's aircraft wandering all over the scenery :shock:

i don't know that i've seen that... it is quite possible that some memory corruption is taking place and overwriting some parts of the data... what i call "buffer overruns" where data is written to a destination without checking there's enough space in the destination...
"You get more air close to the ground," said Angalo. "I read that in a book. You get lots of air low down, and not much when you go up."
"Why not?" said Gurder.
"Dunno. It's frightened of heights, I guess."
User avatar
wkitty42
 
Posts: 5634
Joined: Fri Feb 20, 2015 3:46 pm
Location: central NC, USA
Callsign: wk42
Version: git next
OS: Kubuntu 14.04.5

Re: EDDF ground net

Postby wkitty42 » Thu Apr 12, 2018 5:35 pm

this is working a lot better now but there's still some firm booting i'm having to do... i don't think it is associated with the ground net but i may be wrong on that since i don't know all the details of the ground net's layout and the different points that make it up... it is definitely better than it was with that old parking.xml in place... i started this test round at ** Start Date: 2018-04-12 10:42:15 -0400 **
and it is now 13:33:00 -0400... we've run longer than we have previously...

so... apparently we've also got to make sure there are no old files laying around in any of the old directories that we're not using any more... i wonder if it would be beneficial to clear those distribution directories to remove those old files and then the directories when they are empty? i know we've talked about other similar files in the past and i think they were moved to their new locations or removed from distribution since they were no longer needed...
"You get more air close to the ground," said Angalo. "I read that in a book. You get lots of air low down, and not much when you go up."
"Why not?" said Gurder.
"Dunno. It's frightened of heights, I guess."
User avatar
wkitty42
 
Posts: 5634
Joined: Fri Feb 20, 2015 3:46 pm
Location: central NC, USA
Callsign: wk42
Version: git next
OS: Kubuntu 14.04.5

Re: EDDF ground net

Postby gooneybird » Thu Apr 12, 2018 6:39 pm

wkitty42 wrote in Thu Apr 12, 2018 2:40 pm:
gooneybird wrote in Thu Apr 12, 2018 2:13 pm:There seems to be no pattern to this blocking. I've been using the Channel Islands for testing while working on the groundnets and sat at Guernsey the last week or so and I've quite often had the first aircraft taxi to the end of the runway and just sit there until I quit and restart FG. On the restart all 12 of the test traffic (1 departing every 5 min.) will usually happily depart on time with no hold ups.

do you mean when they get stuck and need a boot to the backside? yeah, i think that's something in the code... i don't know how it is keeping up with the order of the craft it is controlling... i haven't gotten that far in studying the code, yet...


This was a lone aircraft, no other AI was active at the time so It's definitely something in the code.

gooneybird wrote in Thu Apr 12, 2018 2:13 pm:The same goes for the pushback routes, one minute they work perfectly and the next there's aircraft wandering all over the scenery :shock:

i don't know that i've seen that... it is quite possible that some memory corruption is taking place and overwriting some parts of the data... what i call "buffer overruns" where data is written to a destination without checking there's enough space in the destination...


I think has to do with the way AI follow the taxi route and nodes, I seem to recall this was altered a while ago so that the AI didn't follow the turns so tightly at the nodes, also they seem to go for the nearest node instead of the next node on the taxi route, hence the wandering around, they then often go directly to the runway ignoring any taxi routes.


The parking.xml file in %FGDATA%/AI/Airports/EDDF I believe is my old original one done quite a few years ago but it is using the old gate size and location data so will probably be incorrect for the current version of FG.

Current groundnet.
Image

Old groundnet
Image
gooneybird
 
Posts: 2971
Joined: Sat May 31, 2008 1:57 pm

Re: EDDF ground net

Postby gooneybird » Fri Apr 13, 2018 3:24 pm

gooneybird wrote in Wed Apr 11, 2018 7:16 pm:Thanks guys I'll risk 2018.1 then.


Well I bit the bullet and installed it, I run it twice and it only crashed twice ;)
That was after about 30 min. each time with the UFO chasing AI and collecting Terrasync data.
Got to go through my additional settings with a fine tooth comb I think just to see if I can rule them out.

On the plus side there's a lot of new AI aircraft and liveries - Thanks Ian :D

I seem to recall there was a debug setting that only showed AI errors does anyone know what it was?
gooneybird
 
Posts: 2971
Joined: Sat May 31, 2008 1:57 pm

Re: EDDF ground net

Postby wkitty42 » Fri Apr 13, 2018 5:14 pm

i think "--log-class=ai"... maybe "--log-class=traffic"?

can you have more than one class? how would you write that? "--log-class=ai,traffic"?

currently i'm running with "--log-level=debug" and piping the console output to grep to extract the lines i want... those get tee'd to a separate log file and the screen... my system generates like six or seven different logs for each of my play sessions ;)
"You get more air close to the ground," said Angalo. "I read that in a book. You get lots of air low down, and not much when you go up."
"Why not?" said Gurder.
"Dunno. It's frightened of heights, I guess."
User avatar
wkitty42
 
Posts: 5634
Joined: Fri Feb 20, 2015 3:46 pm
Location: central NC, USA
Callsign: wk42
Version: git next
OS: Kubuntu 14.04.5

Re: EDDF ground net

Postby gooneybird » Fri Apr 13, 2018 7:46 pm

I've already got a headache and trying to understand what you're doing only makes it worse ;)

Ok, tried several things, --log-class=ai works as does --log-class=terrasync but nothing gives me what I want and that is problematic groundnets, the first one does give reams of missing airports but I'm not going to go there :shock:

I tried traffic, scenery and groundnet but had no reaction from the consul.
gooneybird
 
Posts: 2971
Joined: Sat May 31, 2008 1:57 pm

Re: EDDF ground net

Postby wkitty42 » Sat Apr 14, 2018 4:28 pm

that's why i grep my console logs... i've also been adjusting the logging code so that it makes it clearer where the entries are coming from... some of this logging is just not currently available in FG because it only exists in my local copy that's i'm hacking on to add more AI related logging to...


a quick explanation of my methods:

tail is a program that follows a text file and shows you the most recent entries added to the file... i'm using it on the log file...
the output from tail is being piped into grep which is a string search tool... it is outputting only the matching text i'm searching for...
in the end, i have a continually updated screen of log entries that only pertain to what i'm looking for...

i'm not doing the class logging thing because i still want/need all of the data logged... some of the log entries i'm looking for come from different classes, i think...

tail -F runfgfsdebug-console.log | egrep -Hin -e "FGAIFlightPlan|exception|\{atc\}|Traffic manager:|could not"

so with the above command line, i'll only get output that has one of these strings in it...
"FGAIFlightPlan"
"exception"
"\{atc\}"
"Traffic manager:"
"could not"

tee is another tool that simply splits the input and writes it to two places... in my case, it writes it to the screen and to another log file... if you remember the reel-to-reel tape machines that had playback heads beside the recording heads, you could listen to what was recorded immediately after it was written to the tape... tee is kinda like that in that you can see the data as soon as it is written to the log file so you don't have to wait until after the run to go looking through the log...

i forget where the log levels and classes are detailed but the wiki should have them... if not, the command line help should... i just don't know if the command line help details all of them...

as far as problematic ground nets goes, i suspect them when there's an error or exception logged... that's why i'm trying to clarify the logging entries so we'll know which ground net file it is and maybe even which parking space or route that is messed up...
"You get more air close to the ground," said Angalo. "I read that in a book. You get lots of air low down, and not much when you go up."
"Why not?" said Gurder.
"Dunno. It's frightened of heights, I guess."
User avatar
wkitty42
 
Posts: 5634
Joined: Fri Feb 20, 2015 3:46 pm
Location: central NC, USA
Callsign: wk42
Version: git next
OS: Kubuntu 14.04.5

Re: EDDF ground net

Postby gooneybird » Sat Apr 14, 2018 5:43 pm

I'll take your word for it, that's beyond my pay grade ;)

as far as problematic ground nets goes, i suspect them when there's an error or exception logged... that's why i'm trying to clarify the logging entries so we'll know which ground net file it is and maybe even which parking space or route that is messed up...


That's far better than the anonymous messages we get now :D

if you remember the reel-to-reel tape machines that had playback heads beside the recording heads,


That sounds like the WEM Copicat echo unit with a tape loop, a real pain in the bum they were, state of the art at the time but so dated now.
gooneybird
 
Posts: 2971
Joined: Sat May 31, 2008 1:57 pm

Re: EDDF ground net

Postby BecOzIcan » Mon May 14, 2018 5:35 am

@wkitty can you please mark the thread resolved (green Tick Top Right) ? Easier to keep track of messages requiring attention. Cheers. Ian
Current Projects: AI Traffic, Models & Liveries
User avatar
BecOzIcan
 
Posts: 1040
Joined: Tue Oct 04, 2011 10:43 pm
Location: Sydney, NSW, Australia
Version: 2.12
OS: Win10

Re: EDDF ground net

Postby wkitty42 » Mon May 14, 2018 7:42 am

i don't think it has been resolved... i haven't been back to EDDF in a week or two... i've been concentrating on trying to track the flow of the traffic manager and the AI craft routines... i wouldn't even know which message to use the green tick of...

a question was raised about the old parking.xml files, too... it looks like the entire %fgdata%/AI/Airports directory tree can be completely removed if everything is switching to ground nets... i've renamed it in my setup to try to eliminate things that may have been causing problems... in the same manner, i've reduced my AI load by wiping %fgdata%/AI/Traffic except for a copy of the misc/WooningAviation.xml which i've used as a base to create LandmarkAviation at KRDU... while it uses the same required-craft, their callsigns are different...
"You get more air close to the ground," said Angalo. "I read that in a book. You get lots of air low down, and not much when you go up."
"Why not?" said Gurder.
"Dunno. It's frightened of heights, I guess."
User avatar
wkitty42
 
Posts: 5634
Joined: Fri Feb 20, 2015 3:46 pm
Location: central NC, USA
Callsign: wk42
Version: git next
OS: Kubuntu 14.04.5

Re: EDDF ground net

Postby BecOzIcan » Mon May 14, 2018 9:58 am

wkitty42 wrote in Mon May 14, 2018 7:42 am:it looks like the entire %fgdata%/AI/Airports directory tree can be completely removed if everything is switching to ground nets...


Not sure where that came from. Wouldn't you loose your runway preferences (RWYUSE) ? You could try it on your local. Only thing I can confirm I tested successfully is that parking.xml is no longer required for as long as you have a groundnet.

Cheers
Ian
Current Projects: AI Traffic, Models & Liveries
User avatar
BecOzIcan
 
Posts: 1040
Joined: Tue Oct 04, 2011 10:43 pm
Location: Sydney, NSW, Australia
Version: 2.12
OS: Win10

Re: EDDF ground net

Postby BecOzIcan » Tue May 15, 2018 12:56 am

wkitty42 wrote in Mon May 14, 2018 7:42 am:i wouldn't even know which message to use the green tick of...


Tick on Top RHS of your first post for this thread. Showed only to the person who created the thread AFAICS

Image

Cheers
Ian
Current Projects: AI Traffic, Models & Liveries
User avatar
BecOzIcan
 
Posts: 1040
Joined: Tue Oct 04, 2011 10:43 pm
Location: Sydney, NSW, Australia
Version: 2.12
OS: Win10

PreviousNext

Return to AI Traffic

Who is online

Users browsing this forum: No registered users and 4 guests