Moving Child Prims on Attachments Relative to Root

[Innula Zenovka]

I was trying to help someone with this in-world, using llSetLinkPrimitiveParamsFast, and, since I know it's been discussed several times in the past here (or in previous versions of this forum), I tried to find a thread to which to direct him, and I couldn't.

Can anyone suggest a thread I might bookmark for future reference?

[wrable Amat]

I do not kno any threads but I do play around a lot with this function.

 

Ask your freind to post the code here and others can help him or her out.I learn how it all works using SL wiki and a lot of time on my hands.

 

[Innula Zenovka]

What we were trying to do was figure out the equivalent of this

integer toggle;vector offset =<0.0,0.0,0.25>;default{    state_entry()    {       // llSay(0, "Hello, Avatar!");    }    touch_start(integer total_number)    {       rotation r = llList2Rot(llGetPrimitiveParams([PRIM_ROT_LOCAL]),0);       toggle =!toggle;       if(toggle){           llSetPos(llGetLocalPos()+offset*r);        }        else{            llSetPos(llGetLocalPos()-offset*r);        }    }}

 only called from a single script in the attachment's root rather than by scripts in the children, and without hard-coding the start pos and end pos.

That is, we wanted to move the child prim along its own local z axis.

 

[Void Singer]

without condensing...

integer  vIntLnk = 2;                //-- child prim link numbervector   vPosOfs = <0.0, 0.0, 0.25>; //--local offset to use //-- Gets on PRIM_POS are always region relative //-- (child pos - root pos) / root rot == relative to root.vector   vPosLcl = (llList2Vector( llGetLinkPrimitiveParams( vIntLnk, [PRIM_POS] ), 0 ) - llGetLocalPos()) / llGetLocalRot();rotation vRotLcl = llList2Rot( llGetPrimitiveParams( vIntLnk, [PRIM_ROT_LOCAL] ), 0 );//-- Sets on PRIM_POS are local to the next highest frame of reference //-- that means child relative to root, and root relative to region.llSetLinkPrimitiveParamsFast( vIntLnk, [PRIM_POS, vPosLcl + vPosOfs * vRotLcl] );//-- local pos + (local offset * local rot) == local relative change in position

 condensed:

integer  vIntLnk = 2;                //-- child prim link numbervector   vPosOfs = <0.0, 0.0, 0.25>; //--local offset to usellSetLinkPrimitiveParamsFast( vIntLnk, [PRIM_POS, ((llList2Vector( llGetLinkPrimitiveParams( vIntLnk, [PRIM_POS] ), 0 ) -  llGetLocalPos()) / llGetLocalRot()) + (vPosOfs * llList2Rot( llGetPrimitiveParams( vIntLnk, [PRIM_ROT_LOCAL] ), 0 ));

 

[Innula Zenovka]

Thank you so much, Void.  That clarifies things hugely.   I was forgetting about offsetting for llGetLocalRot().

[Jake Oxidor]

While this is working fine for me in world, as soon as I attach it, the object stops working alltogether.  I am wondering if one of the functions is not available via attachment or some other caveat that is stopping this from running while in attachment form.

 

-Jake Oxidor, the person Inulla was trying to help.

[Void Singer]

it's still working, it's just that the viewer isn't getting an update for small changes when attached... throw in something like

llSetText( " ", ZERO_VECTOR, 0.0 );

after the call and an update will be forced

[Jake Oxidor]

I added the call for a settext with no text....Still no change when attached.  The offset is .25 still....is that too much or too little for an attachment perhaps?

[Void Singer]

not "no" text, something different than it was... I used a space above, but any character should work (alhtough I haven't needed it recently so the server may cull line breaks/spaces)

[Innula Zenovka]

It's not that, Void.  When it's on the ground, llGetLocalPos returns the position relative to the region if called in the root prim, I discover from the wiki, but when it's attached, it returns the position relative to the attachment point.

So it's failing when attached because it's trying to move the child prim vast distances (I've just verified this with a couple of llOwnerSays).   And I can't figure out how to calculate the properly corrected offset.

[Void Singer]

that's the right thing for it to return, are you saying that llGLPP is returning something different than the child's distance from the attach point?

[Jake Oxidor]

Innula as I play with it I get varying results its either way off like you said or it doesnt move atall.  

[Innula Zenovka]

Yes, on its own, llGetLocalPrimitiveParams(2,[PRIM_POSITION]) returns the sim-relative position in both cases.

[Jake Oxidor]

Ok, I have it finally working, Void had a few errors in it but I fixed them, still can't get it to work as an attachment though, which is kind-of a huge piece of what I am trying to do.  So yea, if anyone knows the issue which is not allowing this snippet to work on attachments and can fix it....it would be much appreciated.

[Void Singer]

ok weird, my mind is slipping... i just checked versus my resizer code, and I'm using llGetPos, and llGetRot there for the root values, which should both report av values.... but checked it and it works... same formula, different functions.

[Innula Zenovka]

Thanks.  I thought llGetPos and llGetRot report the avatar values.  Maybe they do, but that doesn't matter for the purposes of the calculation?  My geometry is not worth a great deal, I'm afraid.    But yes, I've just tried

vector   vPosOfs = <0.0, 0.0, -0.25>; default{    state_entry()    {       // llSay(0, "Hello, Avatar!");    }    touch_start(integer total_number)    {        vPosOfs.z*=-1.0;       integer  vIntLnk = 2;                //-- child prim link number //--local offset to use //-- Gets on PRIM_POS are always region relative //-- (child pos - root pos) / root rot == relative to root.vector   vPosLcl = (llList2Vector( llGetLinkPrimitiveParams( vIntLnk, [PRIM_POSITION] ), 0 ) - llGetPos()) / llGetRot();vector v = llList2Vector( llGetLinkPrimitiveParams( vIntLnk, [PRIM_POSITION] ), 0 );rotation vRotLcl = llList2Rot( llGetLinkPrimitiveParams( vIntLnk, [PRIM_ROT_LOCAL] ), 0 );//-- Sets on PRIM_POS are local to the next highest frame of reference //-- that means child relative to root, and root relative to region.llSetLinkPrimitiveParamsFast( vIntLnk, [PRIM_POSITION, vPosLcl + vPosOfs * vRotLcl] );//-- local pos + (local offset * local rot) == local relative change in position    }}

 and it works at any angle, attached or unattached.

Thanks again for your help, Void.

[Void Singer]

---

Innula Zenovka wrote:

Thanks.  I thought llGetPos and llGetRot report the avatar values.  Maybe they do, but that doesn't matter for the purposes of the calculation?  My geometry is not worth a great deal, I'm afraid.

---

the simplest way to do it is if all calculations are done in the same frame... but the get and sets for position and rotation have always been rather funky.... converting to one frame of reference from another requires the starting values to all be in the same frame... don't ask me how I forgot that llGPP returns a region position for child prims, I was thinking it returned the attachment position. but returning region frame for both lets you convert correctly to local, which is what the set requires.

[Ferd Frederix]

For attchaments, you need to subtract the root rotation. If unattached, you must subtract the local rotation.  I use this in my prim animators:

 

 

 

rotation calcChildRot(rotation rdeltaRot){ if (llGetAttached()) return rdeltaRot/llGetLocalRot(); else return rdeltaRot/llGetRootRotation();}

 

[Void Singer]

llGetRot will work for both, thanks to the buggy nature