Autonomous CAP and GCI AI fighter script

[TBear]

Hi all,

 

can U conferm me that the crashes i see after 15/30 mins is due to some prob in the new patch and not in the scripts?

[Stonehouse]

That's what we believe. Grimes has indicated that the crashes are not present in his tester's version of DCS as I recall.

[SoW Reddog]

How long does it normally take for fixes from ED for stuff that appears an accidental change or unforeseen consequence? I'd like to look at this script but not much point if it doesn't work with the current stable release version.

[Pikey]

Depending on the issue, and im quoting worst case for the group client bug....over a year. I see visual stuff and game stoppers done in a week. Thats the gods honest truth.

How long does it normally take for fixes from ED for stuff that appears an accidental change or unforeseen consequence? I'd like to look at this script but not much point if it doesn't work with the current stable release version.

[Stonehouse]

Really could not comment sorry. It all depends on the internal ED development and testing cycles and their next patch passing their QA checks. Even though Grimes has said that his tester build didn't crash it doesn't guarantee that the build represented by the next patch will include the relevant fix. That said though, I would think there is a fair chance it will be fixed next patch if whatever the code change required passes internal QA as it is causing a lot of grief. Just my guess and opinion though.............

[SoW Reddog]

Thanks. I appreciate the replies. We are just starting out on a journey with DCS that you guys have trodden already, but it's hard picking from the forum and search the threads which still have relevance and work.

 

From what I'm seeing it does look like ED place MP, mission editing and "environment" down the priority scale but we shall see how it pans out.

[Pikey]

My condolences, picking the story up half way would be a true ballache. It's got more history and idiosyncrasy than the wars DCS can depict. Try to link up with someone that has experience and you may avoid some pitfalls. And remmeber, it's only two more weeks.

Thanks. I appreciate the replies. We are just starting out on a journey with DCS that you guys have trodden already, but it's hard picking from the forum and search the threads which still have relevance and work.

 

From what I'm seeing it does look like ED place MP, mission editing and "environment" down the priority scale but we shall see how it pans out.

[SoW Reddog]

My condolences, picking the story up half way would be a true ballache. It's got more history and idiosyncrasy than the wars DCS can depict. Try to link up with someone that has experience and you may avoid some pitfalls. And remmeber, it's only two more weeks.

 

Ah, the "two more weeks" transcends game boundaries and is a universal constant lol.

 

Yup, I don't want to reinvent the wheel, or waste time struggling with things which don't work right now. One of the benefits with CLOD is its stability, no TF patches for 3 years means plenty of time to fix your own mistakes and get things working but the downside is the core bugs never get addressed. Here we have the reverse. Still, a change is as good as a rest.

 

I shall try to find a suitable mentor. Fancy volunteering? ? :D

[Pikey]

If you want to IM me you can find me in the MOOSE slack channel with other people of a similar background and interest. I'm also in the AutoGFT Discord channel. I think joining Moose club would actually suit you as it has a misison building channel.

[Sierra99]

Let me edit this to make sense...

 

I am using the 476th NTTR map and it loads both Moose AND Mist. When I load lukrop's GCI/CAP script, frame rates plummet. (Tested several times to verify script was cause of frame rate issues.) Ciribobs Range script also uses Mist but it doesn't seem to affect frame rates and works flawlessly.

 

Is there a Moose version of the CAP/GCI script available?

 

Sierra

Edited January 28, 2017 by Sierra99

[Stonehouse]

No lukrop's version is mist based.

 

Make a copy of the script. Then do a global search and replace of the string 'mist' and replace it with '_mist'

 

Then add the following line somewhere near the start say just prior to gcicap = {}

 

local _mist = mist

 

Might give you a boost and might not. It makes mist a local table and therefore local function calls. Cheaper for performance than a global table and GCICAP makes a fair few calls to mist quite often.

 

If you stuff up the search and replace then obviously calls to mist will fail but that's why I said make a copy so it's no great loss.

[Pikey]

Hello again Beamrider!

 

MOOSE is a different kettle of fish and works differently. Whilst there is no "like for like" replacement for GCICAP he does have a very interesting MP variant with less features. Effectively its called AI Balancer. It's features, in short, are that it will take off, patrol and land and clean up AI but only if they are set as a client and only if the client slot is unoccupied. The net result is that if you set 8 clients and take up 3 MP slots, then 5 AI spawn and follow a template route and patrol in a zone. And they dynamically change according to the player slot take up. There are no borders or Intercept takeoffs or linking with AWACS or limits on numbers.

 

I am just about to test OB with GCICAP to see about the crashing and performance as It's been hard to tear myself away from the Viggen and having the ability to go up against a Kusnetsov battle group with some real chance of doing damage with the excellent Rb15 programmable cruise missile from IKEA....

Let me edit this to make sense...

 

I am using the 476th NTTR map and it loads both Moose AND Mist. When I load lukrop's GCI/CAP script, frame rates plummet. (Tested several times to verify script was cause of frame rate issues.) Ciribobs Range script also uses Mist but it doesn't seem to affect frame rates and works flawlessly.

 

Is there a Moose version of the CAP/GCI script available?

 

Sierra

[Pikey]

Hey folks of GCICAP land.

 

I've tested Scooterntuz and SHadowD's previously posted missions that were crashing in OpenBeta and am pleased to say there is no CTD in this version.

 

I will add that the performance is questionably worse, some microstuttering, note that I am testing on an i5 2500K @ 3.9GHz OC and in SP with no clients attached and also Tacview processing and LotATC without connected clients and these missions were pretty testy...the CPU isn't getting lower than 40% utilisation with some background apps but I will add that in my experience this is normal, except the stuttering on terrain panning seems new/different and i couldnt get speed multiplacations very high without some scary stops.

 

But it did not crash for the testing period which was hours for Scooternutz and 27mins for the other mission.

 

happy capping next week.

[71st_AH Rob]

Hey folks of GCICAP land.

 

I've tested Scooterntuz and SHadowD's previously posted missions that were crashing in OpenBeta and am pleased to say there is no CTD in this version.

 

I will add that the performance is questionably worse, some microstuttering

 

My server has been running on the OB since Friday, the mission running through its 12 hour cycle with no crash, but confirmed that performance is not as good. Noticeable lag and short freeze when an AI interceptor was spawned with 20+ in the server.

[Stonehouse]

Did anyone try making mist local? Just to see if it helps anything?

[71st_AH Rob]

Did anyone try making mist local? Just to see if it helps anything?

 

Not yet, will be trying that when I get home from work this evening and will report back.

[Sierra99]

Did anyone try making mist local? Just to see if it helps anything?

 

What do you mean "Making Mist Local"?

 

Sierra

[Sierra99]

Hello again Beamrider!

 

MOOSE is a different kettle of fish and works differently. Whilst there is no "like for like" replacement for GCICAP he does have a very interesting MP variant with less features. Effectively its called AI Balancer.

 

Thanks for the explanation Pikey!

 

Sierra

[Grimes]

What do you mean "Making Mist Local"?

 

Mist is loaded as a global table that is accessible to calls from pretty much anywhere in the mission scripting environment. Local values in lua are always a bit faster. So if you make a local reference it is also another way of making it faster.

[Pikey]

Looks like we need to test for regressions in OA this week instead for GCICAP. Let's hope it's not broken for that.

[Sierra99]

Soooooooooooo This Happened 57 minutes into the mission...3 Times in a row...NTTR 2.0.5.1648. Let me know what other info I can provide.

 

Sierra

Edited February 5, 2017 by Sierra99

[Grimes]

Without looking at the code I can make the healthy assumption that the mission might have the M2000 or other aircraft that allow letters as part of the onboard number scheme. It is relatively new, but I know the M2000 and Spitfire use it.

 

Probably just needs to add a check to handle it differently based on whether the value is a number or a string.

[Sierra99]

Without looking at the code I can make the healthy assumption that the mission might have the M2000 or other aircraft that allow letters as part of the onboard number scheme. It is relatively new, but I know the M2000 and Spitfire use it.

 

Probably just needs to add a check to handle it differently based on whether the value is a number or a string.

 

I don't have any M2000 as CAP or GCI.

 

F-15s and F-18s for CAP and GCI Blue, MiG-21Bis and F-5s for CAP and GCI Red...I think it's funny all three stopped at exactly 57 minutes in...

[Grimes]

The full error shows up in DCS.log and it will always be more useful than a screenshot of the error.

 

Still though it looks to be trying to do math on a string value. So it is possible the value stored in the editor went from a number to a string. I kinda assumed it would only be a string if it had letters in it.

[Stonehouse]

Looks to me (as Grimes thought) that this line is the culprit

 

local onboard_num = template_unit_data.onboard_num - 1

 

in function spawnFighterGroup

 

So theoretically if you ensure that the tail numbers are numeric on the template aircraft it should work??

 

Checking lukrop's profile he hasn't been on here since early December 2016 but probably try sending him a PM to see if he will fix his code to cater for alpha as well as numeric. I assume he is trying to make sure that the aircraft in spawned groups have different sequential tail numbers but not really sure. His code is quite different to the original version of the script.