Ghost velocity

BotFodder wrote:

How often is the "Ghosting out of the world" bug occurring on the server?

Increasing the speed of the floating ghost 10x with no other changes to the code results in the bug happening every single time.

I have recoded the check to see when ghost should end, but now the velocity is wonky, and on long gohsts (ONS) it can miscalculate the destination (though the new movement check accounts for that and compensates, it still shouldn't happen.)

Here's the current state of the code, since I've been posting like a madman all darn day and shared nothin'!

This is INCOMPLETE CODE and not suitable for testing, let alone production.

Changes:
-global variables
-uses PHYS_Hovering during ghost movement
-sanity check with OriginalLocation and RevivePoint being concretely assigned before motion starts
-uses new system to check when ghost should end, search on deltaGhostTravel. The variable itself determines how far Instigator is from RevivePoint; the moment it increases (instead of decreases), player is revived.

Moof wrote:

What happens is on the rare occasion that someone's framerate is low, or something else strange happens like a lag burst, the game can actually "miss" when the ghost is supposed to stop.

I'm not sure that's possible. From my understanding, the client doesn't get to move the pawn at all once it starts ghosting, but rather the server is constantly setting the location.

Maybe I'm wrong about this, but I really dont think the client has that kind of control, otherwise it would be easy for somoene to write a hack and control the ghosting pawn.

Just because that's not the cause though, your description makes a certain amount of sense. Maybe the server skips a tick because it's busy.

Would it be possible to measure the distance between the revive point and the pawn as compared to the original start location, and if the distance is ever greater than the current distance to the pawn, cause them to revive at the revive point imediately?

Like I said though, I haven't seen the ghost forever bug in a long long time. I think Mysterial and I fixed the major problem (which involved a physics model other than walking, swimming, or falling) Is this problem still happening and I just dont know about it?

Moof wrote:

Increasing the speed of the floating ghost 10x with no other changes to the code results in the bug happening every single time.

well that sux.

I wonder with the momentum and a lower tick rate on the server, if they never actually hit the revive point, and instead on the next tick simply move past it.

I dont know if this is the same bug as mentioned, but on the map CTF-Bridge of Fate, I ghosted either in the cliff of the gorge, or just over it, not on the bridge, and either died from being in the cliff or fell to my death. Anyway, that should not have been a valid spawn point.

Flak Monkey wrote:

I dont know if this is the same bug as mentioned, but on the map CTF-Bridge of Fate, I ghosted either in the cliff of the gorge, or just over it, not on the bridge, and either died from being in the cliff or fell to my death. Anyway, that should not have been a valid spawn point.

Question:

Did the wave come to an end while you were mid-ghost?

That we're aware of (although I haven't gotten to it and am not 100% sure it's fixable, though I think what we *are* working on is a step in the right direction).

Did the wave come to an end while you were mid-ghost?

It was sometime earlier this year or last year sometime, so the details of that instance are sketchy at most. Just remembered that it happend. I've had a few ghosting enomolies in my past, just not recently.

BotFodder wrote:

That we're aware of (although I haven't gotten to it and am not 100% sure it's fixable, though I think what we *are* working on is a step in the right direction).

The code changes posted above, though ickily buggy still, MAY handle the midwave problem. I have not yet tested it. (It would handle it by teleporting the player to the revive point and immediately reviving them, though I don't yet know if that would handle the two pawn bug seen in Mysterial's original code.)

It would certainly handle Flak's situation. Consider: right now, ghost can "miss" its stop point; that also means there's a plus/minus range around the intended stop point. There's a remote possiblity that if the stop point was close to a wall and he overshot it, it would instantly crater him when he revived.

The reason I'm so sure this can happen is: during the testing I did with different speeds and physics models, this actually happened once with me, where I cratered halfway in the floor at the bio in Deck 17 while ghosting up from the main lava pit.

After all that, it's back to the drawing board. I can patch up some stuff, and I REALLY like a modified version of Szlat's resurrection-by-healing thing (it could happen as a normal part of ghost even - the first time you die, you ghost, and the second, you freeze and hope for a ress), but ANY changes to the physics of the ghost (Flying to Hovering, as suggested by Dru) cause the flying out of the world bug to happen every time. Then, if I try to patch it, I start to get really bizarre vectors coming out of nowhere, or the ghost autocratering every time. The only thing I didn't get to is using DrawDebugLine(). Maybe next try.

So, I'm giving up for now. Work is getting crazy, and in the time I have I'd rather play instead of spend more hours on this.