Footstep sound repeats on ladders (and rapidy in other situations)

[SmallJoker]

When you hold shift on a ladder - you reached the bottom - then it plays the "footstep" sound again and again.

Low priority

[davisonio]

A similar thing happens with fences, when you try climb over them and/or press shift it repeats the sound.

[paramat]

Footstep-farts happen in many situations.

[Desour]

i cant reproduce the original issue but the footstep sound of the node on which the ladder is placed is played very often if above this node is air and you climb down the ladder

[Fixer-007]

Now it repeats while pressing shift on the sea floor after 14aa990

[paramat]

Add #6356 by lupoDharkael
"If you have free_move enabled and keep pressing Shift (without moving), you can hear the footstep sound in loop. The sound should be heard only when you reach the floor."

[lupoDharkael]

I've discovered a new one:

You can experience the sound loop with free_move activated if you move forward at floor level as you can see in the image (no shilft press needed)

[paramat]

Don't you mean 'fast move'? 'free move' would allow you to pass through the nodes in front.

[ghost]

I suspect a duplicate? Nonetheless, rapidly repeating footstep sounds happen easily in my experience, regardless of fast/fly mode. One easy way to reproduce, I think is with liquid nodes.

…

On Wed, Jan 24, 2018, 3:58 AM Paramat ***@***.***> wrote: Don't you mean 'fast move'? 'free move' would allow you to pass through the nodes in front. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1460 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AWXKYneKr7EM5lhqlNZiVdR71L5tPqy4ks5tNvCdgaJpZM4CLJWH> .

[paramat]

Low priority but also, footstep sound is played twice as fast on fences. Folding #4310 into this.

[HybridDog]

I've noticed that no footsteps are played when walking on the ledge, especially noticeable when jumping onto the nodes. It may be intentional because when the foot touches the ledge it should make no sound or a completely different one than when it touches a planar floor.

[paramat]

Lack of edge footsteps sounds is due to a bugfix i made a while ago, see #5199 The alternative is the worse bug that was fixed.

[SmallJoker]

I did some research and found an explanation for this bug.

1. The sound is played here when there's a new Y collision:
   

   minetest/src/localplayer.cpp

   Lines 420 to 421 in 34f940d

   	if(!result.standing_on_object && !touching_ground_was && touching_ground) {
   	m_client->getEventManager()->put(new SimpleTriggerEvent(MtEvent::PLAYER_REGAIN_GROUND));

2. The player controls are applied once before the client environment step:
   

   minetest/src/client.cpp

   Lines 409 to 412 in a43a4e2

   	player->applyControl(dtime, &m_env);
   	// Step environment
   	m_env.step(dtime);

3. LocalPlayer::move (which does the collisions) is called multiple times per environment step to prevent aliasing problems on thin collision boxes.
   

   minetest/src/clientenvironment.cpp

   Lines 200 to 205 in 34f940d

   	/*
   	Move the lplayer.
   	This also does collision detection.
   	*/
   	lplayer->move(dtime_part, this, position_max_increment,
   	&player_collisions);

4. So it happens that the first LocalPlayer::move call detects ground (Y collision) when moving down, but all other in the same env step don't. In the next loop this repeats and the oscillating touching_ground value causes this sound fart.

Possible solution: Only set touching_ground = false (localplayer.cpp) after the steps when none of the LocalPlayer::move collision results returned a Y collision.

[random-geek]

@SmallJoker Just curiously, why are the results of LocalPlayer::move() inconsistent to begin with?

[SmallJoker]

@random-geek move() calls the collision detection function which then moves the player's position and checks for collisions. This happens in small dtime slices (calculated by velocity for maximal travelled distances). So after a certain dtime slice, a collision is detected -> player is on ground now (if Y collision) and it plays the "ground regain" sound. After that, the velocity is back to 0, thus no collision * -> player is floating in the air again.

As already mentioned above, the possible solution for this would be to check whether the player regained ground in one of the dtime slices and compare it with the last step's value to have a more stable variable than touching_ground.

* There is no gravity because liquids and player controls are applied differently.

[LoneWolfHT]

Was about to make an issue but then I saw this 👍

[GreenXenith]

I can no longer reproduce this (ladder/fence/corner repeated step sound) in latest master.

[SmallJoker]

Bug still persists in current master (4445889). Press sneak/shift here:

[GreenXenith]

That is what I did, and cannot reproduce it.

[Desour]

I can still reproduce, too. @GreenXenith have you made sure that your sound volume is loud enough and that the node below has footstep sounds?

[GreenXenith]

Yes, in fact I used the exact node setup as SmallJoker just to be sure.
The footsteps play whilst descending and then stop once I reach the bottom (sneak still pressed).

[paramat]

A very simple and zero-risk improvement is here #9891 (comment)

See #9891 for further discussion and some alternative proposals.