MakeRunways: KATL

[theit8514]

I'm new to using MakeRunways so apologies if this is a known issue or not the right forum. I am trying to update Pilot2ATC's map of KATL and running into taxiways being drawn across the airport:

I first though this could be an issue with the Navigraph Data or the new KATL from the market, but removing both of those folders from the sim's folders still causes the above taxiways to be added.

Any help with this issue is much appreciated.

(Just launched FS2020 and saw it has an update, so this was tested on the previous version)

Edited November 11, 2020 by theit8514

[Pete Dowson]

You are using an early unsupported version of MakeRunways. That problem was fixed within a couple of days of it first being reported!

Pete

 

[theit8514]

On 11/11/2020 at 4:10 AM, Pete Dowson said:

You are using an early unsupported version of MakeRunways. That problem was fixed within a couple of days of it first being reported!

Pete

 

The copy I have is 5.01 and I tried with 5.02 from the website and the same result happens.

[Pete Dowson]

Strange.  Have you tried with the MSFS update?

Pete

 

 

[theit8514]

Yes, currently running Update 6 (1.10.11.0). I extracted out the KATL lines from r5.csv/t5.csv/runways.txt if needed: https://drive.google.com/file/d/1shR333D_1_dZ-FNWBD-eEd7azh3qOOZz/view?usp=sharing

[Pete Dowson]

33 minutes ago, theit8514 said:

Yes, currently running Update 6 (1.10.11.0). I extracted out the KATL lines from r5.csv/t5.csv/runways.txt if needed: https://drive.google.com/file/d/1shR333D_1_dZ-FNWBD-eEd7azh3qOOZz/view?usp=sharing

Okay. Thanks. Those file parts are useful. I'll check -- but it will be tomorrow or more likely Monday now.

Pete

 

[Pete Dowson]

Can you look at some other airports, please? I've done a quick check on the data being used for the KATL taxiways, and they look correctly extracted from the BGL.

I'm therefore wondering whether it is a specific BGL data problem.

I've got nothing which draws the resulting taxiways -- I might need to rectify that with a little graphics program.

Pete

 

 

[theit8514]

I wrote a d3 visualizer for parsing the csv files and finding any point-to-point segment that exceeds 200m. For example, A4 shows multiple points going from the correct area at the north to the SG taxiway in the south:

• Taxiway A4(line 537) point 9 to next point is 3.1 km
• Taxiway A4(line 537) point 11 to next point is 3.1 km
• Taxiway A4(line 538) point 2 to next point is 3.3 km
• Taxiway A4(line 538) point 3 to next point is 3.2 km

https://jsfiddle.net/theit8514/mdqapvr8/5/ (note, this does not filter the CSV, so don't feed it the entire export 😁)

I'll check some other airports tomorrow.

[Pete Dowson]

5 hours ago, theit8514 said:

I'll check some other airports tomorrow.

I checked the data MakeRwys is using for KATL taxiway points and paths (as listed in the log, Runways.txt) and they closely match what the actual BGL contains -- as far as I've managed to check (not easy with several thousand points).

It is important to note that the code in MakeRwys for handling the texiway data is completely unchanged from before and that works fine for all Flight Sims to date.  The only difference for MSFS is the way it is now extracted, s the BGL format has changed a bit (mostly just new block type codes and additional fields, which I'm not using).

So this is why I need to know if all airports are messed up or just KATL (or just one or two), because the former would mean some sort of error in my extraction whereas the latter would certainly mean an Asobo error in forming the data. (Actually of course that could apply to all airports too). 

I'm pretty sure the latest MakeRwys is by now in use by a number of Pilot2ATC and MSFS users, so it may be significant that so far only your report for KATL has been posted.

This may take a while to get resolved.  Thomas is looking at the possibility of making some specialised test tools.

Pete

 

[theit8514]

Thanks! I did test more airports this morning. Large airports KMCO and KMIA render correctly, and I also tested some of the smaller airports in Georgia and those are fine too. Only KATL with the issues so far. All data and images of the renders are here: https://drive.google.com/drive/folders/1NuOG2R2dER6DGr4c6bcQA9iejopIfnbe

[Pete Dowson]

2 hours ago, theit8514 said:

Thanks! I did test more airports this morning. Large airports KMCO and KMIA render correctly, and I also tested some of the smaller airports in Georgia and those are fine too. Only KATL with the issues so far. 

Okay. I managed to get an MSFS T5.csv file processed by P2A this morning (fiddled it, overwriting my P3Dv5 file with one generated on my MSFS PC).

I get NO taxiways drawn at all at KATL. I don't know how P2A has got that strange drawings for you, but I have found out what the problem is after getting the same for LFPG (first reported on the P2A support forum).

MakeRwys is currently a 32-bit program. Dating as it does from FS98 days that's not really surprising.  In order to produce all those files with the airports all in alphanumeric order, it builds up all of the information in memory. It turns out that some of these MSFS versions have huge numbers of taxipoints -- over 4000 for KATL and over 6000 for LFPG.

It simply doesn't get enough memory. It fails 'gracefully' rather than crashing. If I thought there was ever a danger of running out of the 4Gb 32-bit process memory I would have added appropriate error messages as well as acting gracefully, but this has never arisen before -- no where near, in fact.

I'm working on a 64-bit version, but it isn't easy and very error prone (especially because structures in the assorted BGLs are all still packed as if 32-bit), so it's going to take a while and will need another Beta test period. I'll put an announcement up when a Beta version is ready.

Pete

 

[Pete Dowson]

Unfortunately, whilst moving to 64 bit has dealt with the memory problem causing me not to get any taxiways in certain airports, The problem you illustrated in your pic is still there.

I now need to work out whether it is the data in the BGL which is wrong or my interpretation.

You provided this from your analysis:

Taxiway A4(line 537) point 9 to next point is 3.1 km
Taxiway A4(line 537) point 11 to next point is 3.1 km
Taxiway A4(line 538) point 2 to next point is 3.3 km
Taxiway A4(line 538) point 3 to next point is 3.2 km

I'm trying to understand this . The T5 file lists A4 this:

KATL,A4,25.00,33.650642,-84.428642,1,25.00,33.650631,-84.428688,1,25.00,33.650578,-84.428833,1,25.00,33.650471,-84.428963,1,25.00,33.650341,-84.429039,7,25.00,33.650166,-84.429039,1,25.00,33.649990,-84.428963,1,25.00,33.649879,-84.428825,1,25.00,33.649746,-84.428604,1,25.00,33.649609,-84.428230,1,25.00,33.622086,-84.424515,1,25.00,33.622116,-84.424370,1,25.00,33.649548,-84.426849,1,0.00

KATL,A4,25.00,33.650642,-84.429962,1,25.00,33.650608,-84.429771,1,25.00,33.650528,-84.429611,1,25.00,33.621677,-84.425858,1,25.00,33.649990,-84.428963,1,0.00

I assume your point numbers count from 0, else it makes no sense.  So, just taking the first two pairs:

33.649609,-84.428230  -- 33.622086,-84.424515 = point 3859 to 118
33.622116,-84.424370  -- 33.649548,-84.426849 = point 117 to 62

and, indeed, i can see the big change is location, very noticeable.

Armed with just this I can determine whether the problem is in my extraction from the BGL, or in the BGL itself.  My current feeling it the latter as so many airports are okay.

But that's for tomorrow. Getting late here.

Pete
    

[Pete Dowson]

Well, here are the taxipoints as in the BGL (converted to XML): 

<TaxiwayPoint index="3859" type="NORMAL" lat="33.649608604609966" lon="-84.42822799086571" orientation="FORWARD" />
<TaxiwayPoint index="118" type="NORMAL" lat="33.6220857873559" lon="-84.42451402544975" orientation="FORWARD" />

<TaxiwayPoint index="117" type="NORMAL" lat="33.62211797386408" lon="-84.42436695098877" orientation="FORWARD" />
<TaxiwayPoint index="62" type="NORMAL" lat="33.6495465785265" lon="-84.42684665322304" orientation="FORWARD" />

So MakeRwys is getting those right.

BUT there are no paths listed linking those points! So, there is evidently an error in my code which is linking points which shouldn't be linked!  Seems to only occur with these very large airports (in terms of taxipoints).

So, back to debugging ...

[LATER]
Aha! Found it!  

Taking the first bad pair, this is what it should be:

    <TaxiwayPath start="3859" end="4214" ...

Now before MSFS there was effectively a limit of 4095 taxi points, 1-4095. See that 4214 - 4096 = 118, the taxipoint MakeRwys is using.

Ouch! I'll need to delve into it to see what happened to the other value stored in the same place denoting a runway attachment.

A new Beta of MakeRwys should be available later today.

Pete

 

 

    

[Pete Dowson]

Please try the BETA version:

MakeRwys510BETA.zip    

Pete

 

[theit8514]

Pete, the new version is generating data for LEMD and LFPG, but the taxiways for LFPG and KATL are still have the wrong taxi paths. Comparison of the 502 and 510 versions of t5.csv, KATL's taxiways are generated the same in both versions (files equal in diff tool).

[Pete Dowson]

8 hours ago, theit8514 said:

Pete, the new version is generating data for LEMD and LFPG, but the taxiways for LFPG and KATL are still have the wrong taxi paths. Comparison of the 502 and 510 versions of t5.csv, KATL's taxiways are generated the same in both versions (files equal in diff tool).

But the list of Taxi Points (in Runways.Txt) now definitely includes all of them, beyond #4095 which is where it stopped before.  It looks lie even though it now gets the complete list it still only uses the first 4095 in its linking!

Working on it now ...

Pete

 

[Pete Dowson]

Okay. Silly mistake in my previous fix. Please try BETA2, now downloadable

MakeRwys510BETA2.zip

Pete

 

[theit8514]

Everything is looking good on KATL and LFPG and the import also looks good on Pilot2ATC. Thanks for your help!