[#SVC-1179] (Havok 4) llPushObject applied to avatars is inconsistent

2. Second Life Service - SVC

(Havok 4) llPushObject applied to avatars is inconsistent

Created: 12/Jan/08 12:24 AM Updated: 21/Mar/09 03:29 PM

Return to search

Component/s:

Physics

Affects Version/s:

Havok4 Beta

Fix Version/s:

None

File Attachments:

None

Image Attachments:

1. avatar_push_data.png
(7 kB)

Issue Links:

Duplicate

This issue is original of duplicate:

SVC-1818 New Avatar "friction" breaks content

~~VWR-4490~~ Havok4 - LlPushObject not functioning as expected in attachments

Relates

This issue is related to by:

SVC-2513 Push (llPushObject) not working correctly when applied to an avatar!

SVC-141 Restrict Pushing disallows pushes from objects set to group

SVC-875 (Havok 4) Drop off in force applied by llPushObject seems to drop off differently from Havok 1

SVC-754 (Havok4) Meta-issue: Scripting Issues

Linden Lab Issue ID:	DEV-8885
Linden Lab Internal Branch:	havok4-8

Description

« Hide

I've been testing my transport system again, and everything seems to be working fine for dynamic prims (single and linksets).

Avatars are behaving strangely inconsistently though. The setup is as follows:

I have a volumedetecting prim which on collision, gives the object that falls into it a push calculated to give it a horizontal velocity of exactly 10 m/s. I drop my avatar into it, wearing an object that records the first time the horizontal velocity gets above 1m/s, and tells me that horizontal velocity that it gets. This should always return 10.0 when it hits the pusher and, and with dynamic prims it does, out to 5 decimal places. With avatars it is very strangely inconsistent. This is a sample of my data:

[23:57] test pushes: 9.811794
[23:57] test pushes: 10.002684
[23:57] test pushes: 10.003022
[23:57] test pushes: 9.749023
[23:57] test pushes: 9.749105
[23:58] test pushes: 9.625697
[23:58] test pushes: 9.877037
[23:58] test pushes: 9.940375
[23:58] test pushes: 9.939623
[23:58] test pushes: 10.003778
[23:58] test pushes: 10.003645
[23:58] test pushes: 9.749612
[23:58] test pushes: 9.687874
[23:58] test pushes: 10.005015
[23:58] test pushes: 9.812495
[23:58] test pushes: 9.749736
[23:58] test pushes: 9.687205
[23:58] test pushes: 9.940044
[23:58] test pushes: 9.812558
[23:59] test pushes: 10.004038
[23:59] test pushes: 9.625619
[23:59] test pushes: 9.939420
[23:59] test pushes: 9.750255
[23:59] test pushes: 10.004878
[23:59] test pushes: 10.003636
[23:59] test pushes: 9.625620

The attached image plots some of this collected data (about 120 consecutive tests). There doesn't seem to be a local pattern, but there are very clear bands of velocities that we get, rather than random fluctuations. Quantised energy states?! I have no idea what's going on here!

These all have the avatar dropped from the same position (I'm moving there with llMoveToTarget then falling), and I'm also checking that the energy in the pusher prim isn't getting expended (it is always at 1.0 before the push and something bigger than 0.25 after). The same test done with an avatar sized prim looks like this forever:

[0:01] test velocities object: 9.999998
[0:01] test velocities object: 9.999998
[0:02] test velocities object: 9.999998
[0:02] test velocities object: 9.999998
[0:02] test velocities object: 9.999998

Description

I've been testing my transport system again, and everything seems to be working fine for dynamic prims (single and linksets). Avatars are behaving strangely inconsistently though. The setup is as follows: I have a volumedetecting prim which on collision, gives the object that falls into it a push calculated to give it a horizontal velocity of exactly 10 m/s. I drop my avatar into it, wearing an object that records the first time the horizontal velocity gets above 1m/s, and tells me that horizontal velocity that it gets. This should always return 10.0 when it hits the pusher and, and with dynamic prims it does, out to 5 decimal places. With avatars it is very strangely inconsistent. This is a sample of my data: [23:57] test pushes: 9.811794 [23:57] test pushes: 10.002684 [23:57] test pushes: 10.003022 [23:57] test pushes: 9.749023 [23:57] test pushes: 9.749105 [23:58] test pushes: 9.625697 [23:58] test pushes: 9.877037 [23:58] test pushes: 9.940375 [23:58] test pushes: 9.939623 [23:58] test pushes: 10.003778 [23:58] test pushes: 10.003645 [23:58] test pushes: 9.749612 [23:58] test pushes: 9.687874 [23:58] test pushes: 10.005015 [23:58] test pushes: 9.812495 [23:58] test pushes: 9.749736 [23:58] test pushes: 9.687205 [23:58] test pushes: 9.940044 [23:58] test pushes: 9.812558 [23:59] test pushes: 10.004038 [23:59] test pushes: 9.625619 [23:59] test pushes: 9.939420 [23:59] test pushes: 9.750255 [23:59] test pushes: 10.004878 [23:59] test pushes: 10.003636 [23:59] test pushes: 9.625620 The attached image plots some of this collected data (about 120 consecutive tests). There doesn't seem to be a local pattern, but there are very clear bands of velocities that we get, rather than random fluctuations. Quantised energy states?! I have no idea what's going on here! These all have the avatar dropped from the same position (I'm moving there with llMoveToTarget then falling), and I'm also checking that the energy in the pusher prim isn't getting expended (it is always at 1.0 before the push and something bigger than 0.25 after). The same test done with an avatar sized prim looks like this forever: [0:01] test velocities object: 9.999998 [0:01] test velocities object: 9.999998 [0:02] test velocities object: 9.999998 [0:02] test velocities object: 9.999998 [0:02] test velocities object: 9.999998

Show »

All

Comments

Change History

Sort Order:

Seifert Surface made changes - 12/Jan/08 09:07 AM

Field	Original Value	New Value
Summary	llPushObject applied to avatars is inconsistent	(Havok 4) llPushObject applied to avatars is inconsistent

Harleen Gretzky made changes - 12/Jan/08 07:15 PM

Affects Version/s

Havok4 Beta [ 10171 ]

lindenrobot made changes - 16/Jan/08 09:52 PM

Linden Lab Issue ID

DEV-8885

Seifert Surface made changes - 25/Jan/08 01:22 PM

Link

This issue is related to by SVC-875 [ SVC-875 ]

McCabe Maxsted made changes - 30/Jan/08 11:13 PM

Link

This issue is related to by SVC-754 [ SVC-754 ]

Harleen Gretzky made changes - 02/Feb/08 09:18 PM

Link

This issue is duplicated by ~~VWR-4490~~ [ ~~VWR-4490~~ ]

Lex Neva made changes - 11/Mar/08 12:36 PM

Link

This issue is duplicated by SVC-1818 [ SVC-1818 ]

darling brody made changes - 13/Mar/08 09:11 PM

Comment

[ I have found that llPushObject applied to an avatar from an attachment seems to not work when you use a value that is over some new cut off limit. This will break any existing objects that have been scripted within the old lmits.

I posted a script in the duplicated issue ~~VWR-4490~~ which I will post here too. Under Havok1 you can use llPushObject from an attachment or teleporter prim to move the avatar with resonable accuracy and speed. It now seems that previously working scripts are failing under havok4's new cutoff values.

The following script gives my avatar a good 400m push upwards. Under havok4 nothing happens.

[code]
// this is the snippet of code effected by the havok4 change and not the complete push teleporter script.
// Wear an object (cube 0.5x0.5x0.5) containing this script and then touch the object.

PushToTarget(vector DestPos)
{
    key target_key = llGetOwner();
    vector start_pos = llGetPos();
    vector direction = llVecNorm(DestPos - start_pos);
    vector impulse = direction;
    impulse *= llGetObjectMass(target_key);
    llPushObject(target_key, impulse *= llPow(llVecDist(DestPos, start_pos), 3.0), ZERO_VECTOR, FALSE);
}

default
{
    state_entry()
    {
    }

    touch_start(integer total_number)
    {
        PushToTarget(llGetPos() + <0,0,2000>); // destiantion to be pushed to.
    }
}
[/code]
]

Keiko Rau made changes - 02/Apr/08 06:47 PM

Link

This issue is duplicated by SVC-141 [ SVC-141 ]

Keiko Rau made changes - 02/Apr/08 06:48 PM

Link

This issue is duplicated by SVC-141 [ SVC-141 ]

Keiko Rau made changes - 02/Apr/08 06:48 PM

Link

This issue is related to by SVC-141 [ SVC-141 ]

Harleen Gretzky made changes - 23/Jun/08 12:29 PM

Link

This issue is related to by SVC-2513 [ SVC-2513 ]

Andrew Linden made changes - 01/Jul/08 11:18 AM

Assignee

Andrew Linden [ Andrew Linden ]

Andrew Linden made changes - 15/Jul/08 10:38 AM

Status

Open [ 1 ]

In Progress [ 3 ]

Andrew Linden made changes - 15/Jul/08 02:36 PM

Status	In Progress [ 3 ]	Fix Pending [ 10001 ]
Linden Lab Internal Branch		havok4-8

Sue Linden made changes - 13/Nov/08 12:06 PM

Workflow

jira-2007-12-22a [ 51124 ]

jira-2008-11-14 [ 81516 ]

Sue Linden made changes - 13/Nov/08 04:34 PM

Workflow

jira-2008-11-14 [ 81516 ]

jira-2008-11-14a [ 88412 ]

Andrew Linden made changes - 19/Feb/09 09:00 PM

Status

Fix Pending [ 10001 ]

Reopened [ 4 ]