Genapts failure on EIDW

[legoboyvdlp]

TerraGear will load an apt.dat up to version 1100, but it does nothing with those newer features.

Create a EIDW.dat file containing only EIDW. Any change?

the script failed this time at:

Code: Select all

EIDW:Close Contour
EIDW:We still have an active linear feature - add the first node to close it
EIDW: LinearFeature::ConvertContour - Creating a contour with 4 nodes
EIDW:adding Linear Anchor node at lon = -6.29085deg, lat = 53.4227deg, elev = 0m
EIDW:adding Linear Anchor node at lon = -6.29092deg, lat = 53.4222deg, elev = 0m
EIDW:adding Linear Anchor node at lon = -6.29011deg, lat = 53.4222deg, elev = 0m
EIDW:adding Linear Anchor node at lon = -6.29003deg, lat = 53.4227deg, elev = 0m
EIDW:Closed COntour : adding last node at lon = -6.29085deg, lat = 53.4227deg, elev = 0m

Note that I recompiled TerraGear between the two runs.

Load your dat file into WED to confirm the file is valid.

Yep, it loads fine!

Remove the items at the point that the process stops to see if it then runs to completion. Then add items back slowly until you identify the item it does not like. Maybe there is a bug in the loader.

I'm working through it to see if I can diagnose it. Thanks for the help!

[legoboyvdlp]

I got an older version from the gateway, fixed a few issues, and genapts is running right now - previously it would have stopped long before now. I'll readd elements from the newer version (will be faster than removing them from the new version) to see which one has the problem.


I am getting this, not sure if it is concerning:

Code: Select all

rpolate value outside of array!!!
WARNING: Attempt to interpolate value outside of array!!!
WARNING: Attempt to interpolate value outside of array!!!
WARNING: Attempt to interpolate value outside of array!!!
WARNING: Attempt to interpolate value outside of array!!!
WARNING: Attempt to interpolate value outside of array!!!
WARNING: Attempt to interpolate value outside of array!!!
WARNING: Attempt to interpolate value outside of array!!!

but the result looks ok:

Code: Select all

EIDW:Finished airport EIDW : parse 0.562000000 : build 16.223000000 : clean 2.658000000 : tesselate 3.258000000
:Genapts finished successfully

[xDraconian]

The below is more of a warning than an error. It's processing contours and encountered a linear feature. Note that is duplicated the first node to convert a linear 4-node element into a polygon 5-node element.

EIDW:Close Contour
EIDW:We still have an active linear feature - add the first node to close it
EIDW: LinearFeature::ConvertContour - Creating a contour with 4 nodes
EIDW:adding Linear Anchor node at lon = -6.29085deg, lat = 53.4227deg, elev = 0m
EIDW:adding Linear Anchor node at lon = -6.29092deg, lat = 53.4222deg, elev = 0m
EIDW:adding Linear Anchor node at lon = -6.29011deg, lat = 53.4222deg, elev = 0m
EIDW:adding Linear Anchor node at lon = -6.29003deg, lat = 53.4227deg, elev = 0m
EIDW:Closed COntour : adding last node at lon = -6.29085deg, lat = 53.4227deg, elev = 0m

WARNING: Attempt to interpolate value outside of array!!!
Yes, that can be a problem with the resulting terrain.
My understanding is that this message appears while it is attempting to determine the average elevation of a group of points.
An overly simple example would be having points #0, #1, and #2. Then the program tries to determine the elevation of point #-1 (negative) or point #4, neither which exist so you see that message.
The processing continues, but your output may be flawed.

[legoboyvdlp]

All looks ok this morning:


I generated only w007n53 and it took 5 hours without including ogr-decode (only genapts and tg-construct!) :O

It appears that TerraGear now creates Jetways...? or perhaps that was in TerraSync already! Noticed the same at EGAC, where I don't remember any other jetways previously.

[xDraconian]

Looks good.
Glad you got it worked out.

5 hours is a long time to process. I think it is likely that TerraGear needs some more tweaking.

[legoboyvdlp]

To be fair, it was a major urban area with a very high OSM road density, so I wasn't overly concerned - but any improvements will indeed be welcomed

Thanks for all thr help and continued support!

I'm no further on solving the mystery as to why the other layout failed, unfortunately.

P.s. I was surprised to see no noticeable seams or errors so far spotted at the tile boundaries - it would seem terragear has improved in this regard or I was just lucky?

[legoboyvdlp]

This is beyond frustrating - I've an apt.dat file with several hundred airports, and many of them are randomly failing. Do I have to

I've tried both SHM's binaries (the newest), Wlbraggs (WS2.0), and BecOzlacan's and none of them work.
the last output is:

Code: Select all

W50:adding Linear Anchor node at lon = -77.1476deg, lat = 39.2446deg, elev = 0m
remove contour dups : original contour has 4 points
remove_contour_dups: start new iteration
 cur is: lon = -77.1475deg, lat = 39.2446deg, elev = 0m next is : lon = -77.1476deg, lat = 39.2447deg, elev = 0m
 cur is: lon = -77.1476deg, lat = 39.2447deg, elev = 0m next is : lon = -77.1476deg, lat = 39.2447deg, elev = 0m
 cur is: lon = -77.1476deg, lat = 39.2447deg, elev = 0m next is : lon = -77.1476deg, lat = 39.2446deg, elev = 0m
 cur is last point: lon = -77.1476deg, lat = 39.2446deg, elev = 0m next is first point: lon = -77.1475deg, lat = 39.2446deg, elev = 0m
remove_contour_dups : after 1 iterations, contour has 4 points
W50:Parsing pavement: 1 0.25 65.0000 Taxiway
W50:Creating Closed Poly with st 1 smoothness 0.25 thexture heading 65 and description Taxiway
W50:Parsing node:  39.24461166 -077.14753376
W50:Parsing node:  39.24463788 -077.14754521
W50:CLOSEDPOLY::ADDNODE : lon = -77.1475deg, lat = 39.2446deg, elev = 0m
W50:   Adding node lon = -77.1475deg, lat = 39.2446deg, elev = 0m to current linear feature 00000000
W50:Parsing node:  39.24461738 -077.14760958
W50:CLOSEDPOLY::ADDNODE : lon = -77.1475deg, lat = 39.2446deg, elev = 0m

There doesn't seem to be anything obviously bad in the airport in WED.
The polygon it is parsing is a simple rectangle.
There are no new apt.dat 1130 features that TG cannot parse.

Has anyone any suggestions that won't involve me pulling all my hair out?

[wlbragg]

This isn't going to be much help, but I remember similar issues when doing many or even multiple airports. The way I passed the issue was to identify which airports then systematically eliminate portions of the offending airport code that caused the failure. I never analyzed what it was that was causing it to fail, only the chunk of code that caused it.
I remember in at least one instance it was a piece of taxiway or something simple it simply didn't like.

[legoboyvdlp]

Oh dear

Oh well, it has to be done ...

Would be nice if it would skip bad airports instead of quitting.

[xDraconian]

I've tried both SHM's binaries (the newest), Wlbraggs (WS2.0), and BecOzlacan's and none of them work.

How current are these binaries? Were they built with the current TG code on 'next'?

[legoboyvdlp]

SHM's is from 2017, Wlbragg's 2015 (I think), and BecOzlacan's is the oldest.

I've never been able to compile terragear yet owing to link errors. Once I find some spare time I will keep trying.

[Alant]

xDraconians´s windows script fg-from-scratch.cmd , which can be found in the fgmeta git repo https://git.code.sf.net/p/flightgear/fgmeta is working again. It builds simgear, flightgear,terragear and all of their dependencies from the latest git versions. The only problem is that it takes a very long time - about half a day.
All you need is to have Visual Studio and Qt already installed on your machine.

Alan

[legoboyvdlp]

Thanks - I tried it (I already have FlightGear, Simgear, and OSG done, so it was rather quick) and surprise, it worked first go!

So, thanks for the suggestion - I had thought that on my machine it wasn't working but it finally worked for once!




Edit - and a few seconds later - it successfully built airport W50. So looks like its fixed
Thank you very much indeed for the suggestion ... it will save my hair from being pulled out!


Edit - it seems that no dlls have been copied to /bin/ - is this normal that I have to manually copy them from vcpkg?

[legoboyvdlp]

Just an update, I saw something rather odd.
I was running "generate all airports" on my .dat file, when I noticed memory starting to spike. I went away to eat something with memory at 3GB and came back to find 7GB memory useage. The last log output is this:

Code: Select all

   97.32 [INFO]:general    KBWI:Found airport KBWI at 2273506

   97.32 [ALRT]:general    KBWI:

*******************************************************************

   97.32 [ALRT]:general    KBWI:Start airport KBWI at 2273506: start time Thu May 30 12:23:25 2019

Then, it just sat for twenty minutes like that with massively increasing memory useage.

I am running without --threads. I will try again with --threads. I don't think theres anything special about KBWI - in fact, KIAD that it did successfully was more complex!

Edit - I stand corrected; it is something about KBWI, because even using --airport=KBWI, I get the same memory useage.

[legoboyvdlp]

Just a note - I think the memory fault was due to invalid lighting (11.30 features - the new unidirectional taxiway lights) that could not be parsed. WED does not seem to give any error when I try to export 11.30 features to 1000 apt.dat, so perhaps that is more an omission on their end than TerraGear's.