No joke :)

[Smirftsch]

Leo, if the patch is working for ugold too (and i'll try to realize it, whatever it takes) there will be probably one menu for both versions, so these points are good to know to take care about.

[Leo T_C_K]

Umm, can anybody send you umenu file from UnrealGold? It is bad to delete such a great menu. If you don't want or can't fix things at ugold menu, please let umenu be for gold players.

[Smirftsch]

i have the menu, and to be honest, i HATE it, it simply sux in my opinion- but that isnt the point, i donno yet which menu will be used, if it'll be the UT Style Menu or the classic- probably both - if possible, the question is not what i or we prefer, its currently only "what is possible".

[TCP_Wolf]

Even IF the UT style menu is brought to work, you will still need the classic menu for mods. Infiltration uses its own menu, RealCTF does too, so do other mods. Disable the old menu and you disable a LOT of features in mods

[Smirftsch]

i really dont want to disable anything...doesnt matter which menu, i hope i can make them both working to satisfy erveryone...

[Leo T_C_K]

Yeah for realctf menu, infiltration menu, uteamfix menu I binded special key. I can't see problem in that.

[Smirftsch]

the problem is, that not everyone is able to do so

[Leo T_C_K]

the problem is, that not everyone is able to do so

Maybe add to patch unreal.ini that binded key?

[Turboman.]

i'm with leo on this, i think its kinda stupid to patch unreal gold back to the original unreal, at least give gold users a choice or not support gold at all.

thats just my opinion though.

but now enough on the unreal vs unrealgold debate, and lets get back to the original topic discussing bugs

[jackrabbit]

what if you made it a virtual 3D enviroment

[Smirftsch]

i'm with leo on this, i think its kinda stupid to patch unreal gold back to the original unreal, at least give gold users a choice or not support gold at all.

thats just my opinion though.

but now enough on the unreal vs unrealgold debate, and lets get back to the original topic discussing bugs

As already said, its not about what i or we prefer, its whats possible, and as already explained i DON'T HAVE THE UGOLD SOURCES, so we have to see whats possible, and i think its more logical to take the advantages of a new patch instead of not using it because of a menu. Let's see what i can do, if possible i want to provide both menu's.

But fully agreed, lets go back to the topic now.

[Smirftsch]

asgard, its 100% no unrealscript problem, even when using the 224 classes it doesnt work.

[asgard12000]

Yeah its not a script problem, the ringexplosion7 is coded so that it adapts to the way 226 is handling things. Ive got some more onfo on this, will let you know shortly. Its a native problem related to the chain of events.

Also the reason postnetbeginplay cant be used is because its not in the actor class, its in the pawn class.
Later in UT script, they added it to the actor class. I didnt notice before, I guess thats what happens when you code across 2 engines.. doh!

Also there is no problem in the native side or the texture class, well nothing thats relevant here.

I thought at first you you summoned a ringexplosion2 that little purple texture could possibly belong to the server (which can happen sometimes with messy replication) . Were not looking for normal things here.

But thats not the case, its the clients ver were seeing, so that can at least be ruled out.

cheers

[asgard12000]

Well, I wasnt totally correct in saying that its animating on the client. It starts to animate and then stops, so its not a matter of why the client cant see it animate. Fact is that its NOT animating. I guess this is where logs can be deceptive. I did some further logging to check

Or rather it plays the first frame and then is halted.

Here is a 224 server's clients log. I used tick to try and see more.

Code: Select all

ScriptLog:  PostBeginPlay IsAnimating is.......true
ScriptLog:  PostBeginPlay AnimSequence is.......Explosion
ScriptLog:  PostBeginPlay AnimFrame is.......0.001000
ScriptLog:  PostBeginPlay AnimRate is.......1.000000
ScriptLog:  PostBeginPlay AnimLast is.......0.833333
ScriptLog:  PostBeginPlay mesh is.......UnrealShare.Ringex

ScriptLog:  Tick IsAnimating is.......true
ScriptLog:  Tick AnimFrame is.......0.013962

ScriptLog:  simulated state IsAnimating is.......true
ScriptLog:  simulated state AnimFrame is.......0.013962

ScriptLog:  Tick IsAnimating is.......true
ScriptLog:  Tick AnimFrame is.......0.037954
ScriptLog:  Tick IsAnimating is.......true
ScriptLog:  Tick AnimFrame is.......0.051013
ScriptLog:  Tick IsAnimating is.......true
ScriptLog:  Tick AnimFrame is.......0.063715
ScriptLog:  Tick IsAnimating is.......true


MORE ScriptLog etc etc (edited to shorten)

ScriptLog:  Tick AnimFrame is.......0.809610
ScriptLog:  Tick IsAnimating is.......true
ScriptLog:  Tick AnimFrame is.......0.824118
ScriptLog:  AnimEnd.............................
ScriptLog:  Tick IsAnimating is.......false

You'll notice here that it starts playing in Postbeginplay with animframe being where it currently is in the animation sequence. It is still playing when tick has started and further continues past the initial state has started and tick is still logging it beyond.

Eventually the animation completes and thats that.. All good here.

[BOZO]

Maybe check to see if

var bool bAnimByOwner; // Animation dictated by owner.

defined in the Actor class is set by default differently in each 224, 225, or 226 version.


Also try adding a check (in tick or timer) to see if it thinks it is animating on the client. You may want to extend the animation rate to several seconds that way you know all replication is completed and the actor is torn off before you run this check.

native(282) final function bool IsAnimating();


You can also add this function in as another check of if animation ever got called but for some reason but the client is not seeing it because of some native replication issue stemming from when the PlayAnim call was made.

native(261) final latent function FinishAnim();


This is a very interesting problem.

[asgard12000]

Part2 of message

Here is a clients log of a 226 server

Code: Select all

ScriptLog:  PostBeginPlay IsAnimating is.......true
ScriptLog:  PostBeginPlay AnimSequence is.......Explosion
ScriptLog:  PostBeginPlay AnimFrame is.......0.001000
ScriptLog:  PostBeginPlay AnimRate is.......1.000000
ScriptLog:  PostBeginPlay AnimLast is.......0.833333
ScriptLog:  PostBeginPlay mesh is.......UnrealShare.Ringex
ScriptLog:  Tick IsAnimating is.......false
ScriptLog:  Tick AnimFrame is.......0.000000
ScriptLog:  simulated state IsAnimating is.......false
ScriptLog:  simulated state AnimFrame is.......0.000000
ScriptLog:  Tick IsAnimating is.......false
ScriptLog:  Tick AnimFrame is.......0.000000

Like on a 224 server the clients animation has started playing in Postbeginplay. But by the end of postbeginplay it has stopped. Notice tick

Well this is all kind of obvious when you think about it as you see it happen in the game.

The interesting thing here tho is that it seems to be happening at the exact time the actor is initialized.

The reason the ringexplosion7 works is because it doesnt start animating until *after* the actor is initialized.

During the initialization period the animation is being reset or broken somehow.

Is the playanim() function supposed to pass on its variables as it plays? So that when it initializes it has new defaults. (such as whats showing in the 224 log)
with aniframe, animrate, animequence etc etc and its not?

[asgard12000]

Maybe check to see if

var bool                          bAnimByOwner;       // Animation dictated by owner.

defined in the Actor class is set by default differently in each 224, 225, or 226 version.


Also try adding a check (in tick or timer) to see if it thinks it is animating on the client. You may want to extend the animation rate to several seconds that way you know all replication is completed and the actor is torn off before you run this check.

native(282) final function bool IsAnimating();


You can also add this function in as another check of if animation ever got called but for some reason but the client is not seeing it because of some native replication issue stemming from when the PlayAnim call was made.

native(261) final latent function FinishAnim();


This is a very interesting problem.

Just a note these logs are the clients ver of the actor *not* the servers

Well animbyowner needs an owner to work, there is no owner here.
But yeah I did try extending the animation, even tried loopanim() and set the lifespan to 100

The native(282) final function bool IsAnimating(); is how im logging this here or one of the things im logging.

Not sure what you mean to do with FinishAnim(); but FinishAnim(); is kind of misleading, it means wait until the current animation has finished before continuing.
If thats what you mean, it still cant be used in postbeginplay, it can only be used in state code
The problem here is happening before any initial state has started executing code.

But the reason the clients not seeing is is because its not animating, you only see the first frame then its cut, stopped, kaput. It started to animate and for some reason has stopped.

It could be a native replication issue, if the replication hasnt been torn off, eg when the actor is initialized the server tells it the animsequence='' or bisanimating =false or some such
By the time postbeginplay has finished bnettempory actors shouldnt receive any more replication so anything you place in postbeginplay overrides what the server initially sent. They are replicated, and receive no further updates. But then is working correct

Also when playanim is executed on the client it should be saying this is the animsequence,animframe, bisanimating = true etc etc so perhaps its not over riding what was initially sent by the server or is just failing.

I dont have any answers, only questions

Cheers

[Smirftsch]

interesting, interesting...will check some things...
explains why i couldnt reproduce it in 224

Asgard, could you send me your logging code ?

[asgard12000]

Ok sent.. Just another note about this. Around the end of postbeginplay in a normal actor is when variables are replicated to it. In our bnettempory actor, I dont believe this should be happening, the replication should of been torn off by now. Im having serious doubts about this and thinking that 226 is tearing off the replication after the initialization period allowing the server to reset the animation.

This is from the wiki links I posted earlier

PostBeginPlay versus PostNetBeginPlay
The difference between these two functions is small, but important. PostBeginPlay is called after the actor is created but BEFORE any values have been replicated to it. etc etc

also altho there is no bTearOff bool in unreal1 but it covers bnettempory

Tearing Off Actors
etc etc
If the variable bNetTemporary is set to true, this actor will be 'torn off' on a given client as soon as it is initially replicated to that client.
etc etc


So far any tests Ive done prove there is a fault here, such as setting the mesh and drawtype in default properties and then changing them in postbeginplay
if ( Level.NetMode != NM_DedicatedServer ) etc etc

But then i dont know for sure how the engine handles this.

cheers

[asgard12000]

Ok here is proof that the server has messed with things such as a late replication call at initializaitin time

In this class Ive set the mesh as Bird1 and animsequence as flight in default properties. Then in postbeginplay ive changed it to ringexplosion and asked it to play the Explosion animation.

Code: Select all

class RingExplosion8 extends RingExplosion2; 

                  
simulated function PostBeginPlay()
{
 
 if ( Level.NetMode != NM_DedicatedServer )
      {
  Mesh=LodMesh'UnrealShare.Ringex';
  PlayAnim( 'Explosion', 0.2 );
            }
            if (level.bHighDetailMode) SpawnEffects();
            PlaySound(ExploSound,,20.0,,1000,0.6);               
}

defaultproperties
{
 Lifespan=2.000000
 Mesh=LodMesh'UnrealShare.Bird'
 animsequence=Flight
}

So the clients ver of this temp actor *should* play the exposion.

On a 224 server the client sees the normal ringexplosion but on a 226 server the client logs it as this

Code: Select all

ScriptLog:  PostBeginPlay IsAnimating is.......true
ScriptLog:  PostBeginPlay AnimSequence is.......Explosion
ScriptLog:  PostBeginPlay AnimFrame is.......-0.166667
ScriptLog:  PostBeginPlay AnimRate is.......1.000000
ScriptLog:  PostBeginPlay AnimLast is.......0.833333
ScriptLog:  PostBeginPlay mesh is.......UnrealShare.Ringex
ScriptLog:  Tick IsAnimating is.......false
ScriptLog:  Tick AnimSequence is.......Flight
ScriptLog:  Tick AnimFrame is.......0.000000
ScriptLog:  Tick mesh is.......UnrealShare.Bird
ScriptLog:  simulated state IsAnimating is.......false
ScriptLog:  simulated state AnimSequence is.......Flight
ScriptLog:  simulated state AnimFrame is.......0.000000
ScriptLog:  Tick IsAnimating is.......false
ScriptLog:  Tick AnimSequence is.......Flight
ScriptLog:  Tick AnimFrame is.......0.000000
ScriptLog:  Tick mesh is.......UnrealShare.Bird

The 226 server has reset the mesh, anim sequence and no doubt everything else.

Cheers

[jackrabbit]

hmmm very very interesting .. that is definitly somthing you guys should be looking into to fix for the 226 server!

[Smirftsch]

thats exactly the problem.

[Smirftsch]

maybe a weird idea, but if i could make the engine crash within this function i may can use the debugger to find out where the problem appears.

[asgard12000]

Well you can always do what they did in UT.

They made a bool for client animation and if its true its not replicated

In the actor class

Code: Select all

class Actor extends Object
      abstract
      native;


var                    bool              bClientAnim;

replication
{
unreliable if( DrawType==DT_Mesh && ((RemoteRole

[Smirftsch]

unfortunately the replication system of ut differs in this point very much, but its maybe another good idea, i think i could make it.