• All submissions to this site are governed by Second Life Project Contribution Agreement. By submitting patches and other information using this site, you acknowledge that you have read, understood, and agreed to those terms.
Issue Details (XML | Word | Printable)

Key: SVC-1541
Type: Bug Bug
Status: Resolved Resolved
Resolution: Fixed
Priority: Showstopper Showstopper
Assignee: Andrew Linden
Reporter: Charlene Trudeau
Votes: 3
Watchers: 3
Operations

If you were logged in you would be able to see more operations.
2. Second Life Service - SVC

Rez-at-root working improperly - does not seem to recognize old order of linkset rules

Created: 13/Feb/08 03:38 PM   Updated: 01/Apr/08 03:28 PM
Return to search
Component/s: Physics, Scripts
Affects Version/s: Havok4 Beta
Fix Version/s: 1.20.0 Server

Environment:
Second Life 1.19.0 (79674) Feb 8 2008 16:51:12 (Second Life WindLight)

You are at 219235.5, 296753.7, 37.8 in SkyBeam located at sim5578.agni.lindenlab.com (8.2.34.134:13006)
Havok4 Beta Server 1.19.0.79818

CPU: Intel Pentium 4 (Unknown model) (2992 MHz)
Memory: 2046 MB
OS Version: Microsoft Windows XP Service Pack 2 (Build 2600)
Graphics Card Vendor: ATI Technologies Inc.
Graphics Card: Radeon X1900 Series (Omega 3.8.442)
OpenGL Version: 2.1.7169 Release
LLMozLib Version: 1.1.0 (Mozilla GRE:1.8.0.12_0000000000)
Packets Lost: 1286/249241 (0.5%)
Viewer Digest: 3399391e-5c2d-78b2-721e-f28dc9172533
Issue Links:
Duplicate
 

Linden Lab Issue ID: DEV-10586
Linden Lab Internal Branch: havok4-3


 Description  « Hide
This has broken vendors in use by five businesses on SkyBeam.

In trouble shooting the vendor and why it rezzes incorrectly, the following appears to be happening:

It does not recognize the old linkset order. In ones already in the vendor it is apparently using the linkset that it the overall object is named (the last linkset selected prior to 'take'). In new linksets taken, in preliminary testing I have not found any pattern to which one it uses to name the linkset or to rez.

It rezzes the root prim of whatever linkset it thinks is primary at the correct center, probably at a 0,0,0 rotation as some things rez sort of ok while others end up in the ground.

This is, unfortunately, a show stopper for us personally, and we will be requesting rollback to havok 1 until it can be resolved.



 All   Comments   Change History      Sort Order: Ascending order - Click to sort in descending order
Charlene Trudeau added a comment - 13/Feb/08 09:31 PM
Further testing indicates that it is now using the last linkset chosen to use as the root prim/linkset instead of the first linkset's root prim when it rezzes at root.

This means we can fix it by repacking the rezzer and reversing the order the prims are selected, but only if we know this is intended future behavior expected. Otherwise it is simply an enormous amount of hours wasted to work around a bug. We need info on this ASAP so we can decide how to proceed.


Dominic Webb added a comment - 13/Feb/08 09:56 PM
I just spent some quality time coming up with a repro case for this.

Historically, when using llRezAtRoot() with coalesced objects, the FIRST selected linkset has determined the relative position of all other selected linksets.

Now, it appears that this has changed to the LAST selected linkset, but only on sims running the Havok4 code.

Repro case:

1. Rez any number of prims, and shift-select them.
2. Take into inventory
3. Drop the following script into another prim:

default {
on_rez(integer s){ llResetScript(); }
touch_start(integer total_number)

{ llRezAtRoot(llGetInventoryName(INVENTORY_OBJECT, 0), llGetPos(), ZERO_VECTOR, ZERO_ROTATION, 0); }

}

4. Drop the coalesced object from 1) into prim from 3)

Observation: Whichever prim/linkset was selected LAST rezzes at the center of the rezzing prim.

One wonders if this might be related to the Havok4 fix for SVC-1449.


arya pfeffer added a comment - 13/Feb/08 10:49 PM
My limited testing shows that it changed from first to last item select between 1100 SLT yesterday and 0900 SLT today. Both times havok4 was running as far as I know.

Andrew Linden added a comment - 14/Feb/08 10:59 AM
Updating to Showstopper. This one looks like it breaks content.

Andrew Linden added a comment - 14/Feb/08 02:07 PM
Fixed internally, this will go out in the next update, whenever that is.

Yes, fixing SVC-1449 broke this.


Andrew Linden added a comment - 01/Apr/08 03:28 PM
Fixed and deployed... several weeks ago, if not months.