One plugin/server bug that's still present and hugely annoying is AI planes sitting on the tarmac at their parking and gate with their engines idling for minutes on end, not moving a single inch. Not sure if they get stuck between having been cleared for startup and taxi or if they're too close to the user aircraft and too afraid to move, but this really needs to be fixed because it's grinding into my nerves. Restarting the radio (i.e. resetting the plugin) does not help.

Edited October 13, 20214 yr by Bjoern

1 hour ago, Bjoern said:

hugely annoying is AI planes sitting on the tarmac at their parking and gate with their engines idling for minutes on end

Hugely annoying definitely a thing, right now the timings should be about right

airports can also get quite busy (love how they cut this at 3 mins 50)

vs

Big airports are a less less fun when there is traffic in the way.... 😞

whats missing for me is there is no pushback animations or ground crew, and the engine sounds begin as soon as they get startup permission.

what actually irks me most is the XP ground cars driving through everything - not least because I dont think there is anything I can do to fix that because its an XP bug.

Plan is still to replace a lot of that with instance scripting in XTLua (so add ground crew, customise local behaviour based on state - i.e. moving backwards with no engine sound, probably a follow me car), great news there is all that openAL headache can go in the bin because for XP12 we can attach .snd files to instanced objs, that should go in to the plugin almost as soon as XP12 is out to devs.

1 hour ago, Bjoern said:

Restarting the radio (i.e. resetting the plugin) does not help.

Pros and Cons of a persistent world... once crashing into planes or shooting them with guns kills them that should improve.

 

Edited October 13, 20214 yr by mSparks

14 hours ago, mSparks said:

Hugely annoying definitely a thing, right now the timings should be about right

If they are, they're not on the end user side yet.

To be fair, a huge part of the annoyance stems from the wav file used for the sounds. Maybe I can find a better one.

 

14 hours ago, mSparks said:

Big airports are a less less fun when there is traffic in the way.... 😞

Does AutoATC's traffic react to the user aircraft if I don't have a flight registered in the app?

 

14 hours ago, mSparks said:

whats missing for me is there is no pushback animations or ground crew, and the engine sounds begin as soon as they get startup permission.

You'd have to define points of interest (nose gear, L1 door, baggage doors, etc.) for each AI aircraft model used by AutoATC and keep that in a database, but I don't see why this can't be implemented.

The question is whether the additional drawcalls and constant instance and object spawning/despawning is tolerated by users on weaker PCs.

Engine startup/shutdown can also be implemented with the right sounds and delays before getting moving.

 

14 hours ago, mSparks said:

what actually irks me most is the XP ground cars driving through everything - not least because I dont think there is anything I can do to fix that because its an XP bug.

You wouldn't want those vehicles constantly checking for collisions as every CPU cycle is precious.

 

14 hours ago, mSparks said:

Plan is still to replace a lot of that with instance scripting in XTLua (so add ground crew, customise local behaviour based on state - i.e. moving backwards with no engine sound, probably a follow me car), great news there is all that openAL headache can go in the bin because for XP12 we can attach .snd files to instanced objs, that should go in to the plugin almost as soon as XP12 is out to devs.

That's a good improvement for more engine noise variety. I assume spatial positioning akin to FMOD will also be considered?

1 hour ago, Bjoern said:

Does AutoATC's traffic react to the user aircraft if I don't have a flight registered in the app?

yes, at least for taxiing, when i get round to resolution advisories in the 744 the ai planes will use the same logic.

1 hour ago, Bjoern said:

You wouldn't want those vehicles constantly checking for collisions as every CPU cycle is precious.

This simply isnt the case, there has been a lot of lame excuses thrown around for not having collision detection in xplane, but its been a solved problem for more than a decade.

1 hour ago, Bjoern said:

That's a good improvement for more engine noise variety. I assume spatial positioning akin to FMOD will also be considered?

by the looks of it the same tech they have just demonstrated for the ground traffic and tree sounds in the recent keynote.

1 hour ago, Bjoern said:

The question is whether the additional drawcalls and constant instance and object spawning/despawning is tolerated by users on weaker PCs.

yep, proof of concept first. Which will simulate chasing lots of rabbits around, just waiting on the rabbit animation.

Edited October 14, 20214 yr by mSparks

23 hours ago, mSparks said:

yes, at least for taxiing, when i get round to resolution advisories in the 744 the ai planes will use the same logic.

RA only apply while airborne though.

 

23 hours ago, mSparks said:

This simply isnt the case, there has been a lot of lame excuses thrown around for not having collision detection in xplane, but its been a solved problem for more than a decade.

Physx is a proprietary library and Laminar does not like proprietary libraries. And Teardown works on a much smaller scale than a big airport.

 

P.S: If you like your destructions, Red Faction Guerilla is pretty much a classic example of how to engagingly gamify it.

3 hours ago, Bjoern said:

Physx is a proprietary library and Laminar does not like proprietary libraries.

But thats a very different argument to "it can't be done because cpu cycles" i was using it to show why that isn't remotely true.

Enhanced Cloudscapes/Skyscapes is colliding rays with cloud volumes and storing the results across 200kmx200km+ millions upon millions of times a second, and they have better performance than default clouds. Even on the CPU side good results can be had with cheap radius comparisons.

They even have a fully functional ground probe in the SDK - that hasn't been updated for probably a decade.

There is only one reason they don't have collision detection, they never put dev time on it, absolutely nothing to do with any technical difficulties or limitations, even fsx allowed tree and power cable collisions.

3 hours ago, Bjoern said:

RA only apply while airborne though.

yes, on the ground they already wait, (and have done since 2018, although there have been a lot of refinements since then):

in the air they just crash (with crashing currently disabled), when they should take avoidance measures.

 

 

.

 

Edited October 15, 20214 yr by mSparks

Anyway, back on topic, actually flight testing.

"For Landing" request now works, that one had supposedly worked for a while, but I found it was being auto corrected to "fower landing" or recognised as "full landing" which then didn't get processed. Autocorrect added to the android apps (go into ground school to get the update)

I should have now fixed taxi instruction naming across runways - lima/romeo/charlie is now left/right/center, that one threw me until I realised it was giving runway names.

Also rolled out more of the "ticketing" into the actual language logic, this time expiring taxi tickets once you are ready to take off or airborne so it doesn't keep demanding an answer to a failed taxi read back once you are airborne (I told the AI to go f itself when that happened to me, it didn't seem to care).

Most importantly, I finally got around to making a parser for ATCLog files (flight/ATC "replay" gets stored in plugins/AutoATC_java/atclog for a few days)

I can now use those to instantly replay flights to fix behaviour when things act funky (such as getting a tie down parking in a 737, requests/read backs failing etc), I think the best way to deal with this is to attach the one from the flight (they are simply text files you can read with notepad) to a github issue when something doesn't work as expected.

 

 

2 hours ago, mSparks said:

I told the AI to go f itself when that happened to me, it didn't seem to care

Been there, done that.

 

P.S:

Without the app connected, MaryTTS-generated ATIS overlaps X-Plane ATIS. I've added a comment with a suggested fix to this issue.

1 hour ago, Bjoern said:

I've added a comment with a suggested fix to this issue.

woudln't it make more sense to "just" disable maryTTS ATIS voices if xplanes atis_enabled is true?

Todays testing - went well, even (eventually) managed to read back the taxi instructions at EGCC, basic conclusion "needs a follow me car", there's one in 

https://csl.x-air.ru/info/item/150

I'll add the CSL definition to the plugin and add a follow me car AI class asap.

Edited October 18, 20214 yr by mSparks

5 hours ago, mSparks said:

woudln't it make more sense to "just" disable maryTTS ATIS voices if xplanes atis_enabled is true?

Testing yielded that "atis_enabled" is always true unless you set it to false via dataref. But yes, you can basically do that as long as you account for the following scenarios:

1. Plugin, no app, "atis_enabled" true --> X-Plane ATIS
2. Plugin, no app, "atis_enabled" false --> MaryTTS ATIS
3. Plugin, app, "atis_enabled" true --> ???
4. Plugin, app, "atis_enabled" false --> App ATIS

Maybe add a checkbox that governs ATIS usage to the app, i.e. "Use X-Plane ATIS" to elegantly solve the conflict in scenario 3?

This would suppress app and MaryTTS ATIS output and leave it to X-Plane, with all the associated risks (runway usage discrepancies between XP and AutoATC).

15 minutes ago, Bjoern said:

Testing yielded that "atis_enabled" is always true unless you set it to false via dataref. But yes, you can basically do that as long as you account for the following scenarios:

1. Plugin, no app, "atis_enabled" true --> X-Plane ATIS
2. Plugin, no app, "atis_enabled" false --> MaryTTS ATIS
3. Plugin, app, "atis_enabled" true --> ???
4. Plugin, app, "atis_enabled" false --> App ATIS

Maybe add a checkbox that governs ATIS usage to the app, i.e. "Use X-Plane ATIS" to elegantly solve the conflict in scenario 3?

This would suppress app and MaryTTS ATIS output and leave it to X-Plane, with all the associated risks (runway usage discrepancies between XP and AutoATC).

how does that tie in with sound settings:

23 minutes ago, mSparks said:

how does that tie in with sound settings:

That's what I tested. The dataref does not change regardless of the status of those text boxes or master sound checkbox.

1 hour ago, Bjoern said:

That's what I tested. The dataref does not change regardless of the status of those text boxes or master sound checkbox.

but if you have those off you shouldnt get any atis either?

is atis_enabled not just to disable atis (e.g. for weather injectors) when xplanes atc is enabled?

Those checkboxes have no effect on Linux as output is played regardless (X-Plane bug/XLinSpeak quirk), at least when it comes to ATIS.

Only the dataref reliably kills X-Plane ATIS output.