Fixed Player can turn off pushers in FOFs with pushable object if standing near the FOF

[Ezer'Arch]

This bug is not critical (o well, see EDIT2). I was making some experiences in my maps, then I discovered this:

1) Build a hole and put an intangible FOF in the hole (the blue tint is for better visualization);

2) Give the FOF an upward wind effect (linedef type 542 or 545);

3) Drop a pushable object into the hole, the object will float as intended;

4) Player stands up near the hole, the wind effect is turned off and the object gets stuck in the bottom;

5) Player moves away, pushable object skyrockets ^^

Sample wad: http://dl.dropbox.com/u/4799936/SRB2/Files/UpwardWindBug.wad

EDIT:

Not only standing near, fly over as Tails, the bug persists.

EDIT2:

Tested with a player and 1 bot (it's likely to happen in netgame):

 

[Sergeant Brown]

Maybe it's an OpenGL problem? You should screenshot that in software, just to be sure.

 

[Ricardo]

No, it's not an OpenGL problem, just tested in Software and confirmed it.

 

[Metal-Rawr]

It doesn't really "skyrocket" though. Just about x1.5 the height it originally reaches to.

 

[Ricardo]

All right, I found the problem. =)

In the code below (found in src/p_spec.c)...

Spoiler

if (thing->eflags & MFE_VERTICALFLIP)
{
	if (referrer->floorheight > thing->z + thing->height
		|| referrer->ceilingheight < (thing->z + (thing->height >> 1)))
		return;

	if (thing->z < referrer->floorheight)
		touching = true;

	if (thing->z + (thing->height >> 1) > referrer->floorheight)
		inFOF = true;

}
else
{
	if (referrer->ceilingheight < thing->z || referrer->floorheight > (thing->z + (thing->height >> 1)))
		return;

	if (thing->z + thing->height > referrer->ceilingheight)
		touching = true;

	if (thing->z + (thing->height >> 1) < referrer->ceilingheight)
		inFOF = true;
}

... the two "return"s are the culprits.


NOTES:

• This bug actually affects any kind of pusher, not just up-currents/winds.
• This bug only affects pushers on FOFs and the following explanation of the problem takes that in consideration.

Since the aforementioned block of code is called inside a for loop, iterating over the things touching a pusher sector (touching the target sector), and because when a new thing enters a sector it gets added to the head of its touching list, the next time the thinker T_Pusher is called, if the new thing is above/below the FOF's ceiling/floor, the thinker aborts instead of testing further things.

So, the solution is simply to replace those "return"s with "continue"s.

 

[Callum]

Nice find, Ricardo. ;) I don't always have time during the school week to troubleshoot bugs. Committed the change.

 

[Ezer'Arch]

Thanks Ricardo. :D

 

[Ricardo]

No problem ;)

 

[Ezer'Arch]

Let me question a thing: does this bugfix fix this case too?

Stack two or more pushable objects and they aren't affected by pusher effect, even if the player is away from the FOF (MOB r6664:6668).

Just to make sure if we really got rid of this bug entirely.

EDIT:

In the old times, when I was used to host netgame serves, I had seen a similar bug: if two or more players are in the same waterslide FOF, one of players suddenly stands up for few tics and then goes back into S_PainState.

 

[Ricardo]

Yes Ezer, it fixes that case too.

Also,

SpainState.

this cracked me up. xD

 

[Ezer'Arch]

this cracked me up. xD

XD ... I was intending to write S_PainState.

¡Disculpe! pueblo de España. :3