Havok4 Early Adopter Update - RC0 of the New Second Life Simulator with 15 Fixes - 2008-03-07
Friday, March 7th, 2008 at 6:01 PM by: Sidewinder LindenQuick Status
The Havok4-based Second Life simulator v1.19.1.81747 has been deployed to both the Beta Preview and the over 500 Early Adopter regions. This version resolves 15 of the remaining issues and brings us to the point of having a first release candidate of the simulator. More fixes are still in process - this is not the last build - but we are getting much closer to a version that can be integrated with release and become the next Second Life simulator.
We are getting close to a release-ready version of the new Havok4-based simulator. The team will be continuing to work on issues. I will be posting more information about expected timetables soon (but not with this blog post…)
The team is committed to the quality of this new simulator version, and will be remaining together to work issues after the Havok4 simulator is released to Second Life at large. (We will not practice “release and run” with this simulator, just as you have seen that we do not practice “post and run” on the blog… We’ll still be around ;). We are trying to get the benefits of reduced simulator crash rates out to all of Second Life as soon as it is reasonable, and will clean up lower level issues going forward from that point.
Why Did We Not Deploy This Version To Beta Preview First This Time?
The original plan was to deploy this build to the Beta Preview only, and test it over the weekend before deploying it to the Havok4 Early Adopter regions. As a result of a recent security exploit and the accompanying fix, I decided that it would be better to bypass the Beta Preview period than to leave the exploit “in the wild” with our Early Adopters.
I hope that we have caught any major issues that might be in this build, and that you will understand this decision is based on a security first philosophy, even though it may risk some other problems. If we missed something major, the back-out plan will be to pave over the Havok4 Early Adopter regions with the current Second Life simulator temporarily. I will be on and offline for the next several hours and will be making that call later on tonight (keep this version online or pave it over with the release version).
What Has Changed In This Version?
Changes in this build include (SVC items were submitted on the public jira issue tracker / DEV items are internally discovered issues / there are a few in this list that were fixed without jira tracking #’s…).
*** Resolved security exploit
*** Added the fix for the script performance issue that was picked up through our merge to the 1.19.1 Second Life Simulator. This is the same fix that was deployed to the non Havok4 Second Life simulators with the last rolling restart.
SVC-1680: llGetMass() now returns correct values for child prims after rez without having to recompile the script
SVC-1700: Non-physical child prims of physical root prims now behave correctly with llTargetOmega()
SVC-1469: Items no longer deleted or returned from off-world above 768m after region restart
SVC-1299: Sit targets adjusted to more closely match release simulator placement
SVC-1488: Prim.Docker automatic prim alignment no longer leaves gaps
DEV-10635: Personal Hugger (90 sec) and Lost Dog Hugz Sphere hug mis-alignment corrected
DEV-10702: Push-based weapons are now functional (the previous deploy had code that would penalize attempts to push “overly hard” to the point that many weapons displayed zero push - whereas now they are simply capped to the max push allowed)
SVC-1542: Many concurrent shape changes no longer causes physics to stop being applied to objects (a.k.a. “the bunny problem” 
SVC-1688: Simulator now recognizes changes to child prims without waiting for a change to the root object
SVC-1261: Avatars standing on sloped terrain no longer slide down the slope slowly
SVC-1066: Temp prims no longer count towards region prim limits
DEV-10985: Discrepancy between visual and physical collision location resolved for fast-moving object
DEV-11599: Removed new crash mode discovered in final QA process
Best regards,
Sidewinder Linden
Havok4 Program Manager
March 7th, 2008 at 6:05 PM
Yes! now I can sit like a proper lady once more!
March 7th, 2008 at 6:08 PM
/me has panic attack…
http://jira.secondlife.com/browse/SVC-1683 please…
March 7th, 2008 at 6:13 PM
@2 Aria: Have faith! That bug was being worked on today, and is close to resolution. The fix for it should be in our next deploy
/Sidewinder
March 7th, 2008 at 6:20 PM
well sidewinder…the latest client u made me switch to is not stable…so speak up dude…no hiding with the ‘off topic thing’…u people need to get honest and get honest fast with these comment things…every post by sl needs to be open to comment…and sl needs to get deadly honest in replying comments like this…get it?
March 7th, 2008 at 6:20 PM
Sidewinder — I’m still hoping to get one of my regions added to the early adopter program… I’ve had the support ticket filed for about a week and a half now. When I contacted support they were saying that they ran out of ’slots’ for the test but were hoping for more. Do you know, will there be any more ’slots’ added for the early adopters? Thanks!
March 7th, 2008 at 6:25 PM
I am still looking forward to Proper Prim Alignment when coping hollow prims, using (copy/Wand) Create and copy selection
hope its fixed soon….
March 7th, 2008 at 6:27 PM
@4 Wake up: As you well know, I’m not involved in viewer releases at this point, and *I* did not force an update. I cannot control what other folks do and don’t do with blog posts. Which viewer are you talking about? Release, RC, Windlight? I have been running the latest WIndlight, which on my machine has been much more stable, but stability is often highly dependent on graphics hardware specifics. /Sidewinder
March 7th, 2008 at 6:28 PM
@6 RodneyLee: Is there a jira with this specific issue? /Sidewinder
March 7th, 2008 at 6:33 PM
sidewinder i have been using windlight for some time…but today sl forced some kind of upgrade to a client that is totally unsatisfactory to what i was using…send me an im and i will try to help you (sl)
March 7th, 2008 at 6:35 PM
[...] Linden Lab just announced that Havok4 simulator code has now been promoted to Release Candidate 0. [...]
March 7th, 2008 at 6:41 PM
[...] Linden Lab just announced that Havok4 simulator code has now been promoted to Release Candidate 0. [...]
March 7th, 2008 at 6:42 PM
@10 Comment re Daikon Forge: Please see notes above regarding the fact that we are not yet done with fixes for the public release. Best Regards, Sidewinder
March 7th, 2008 at 6:44 PM
@Sidewinder - Understood. Actually, feel free to delete that pingback, as my post was basically written before I had a chance to fully think about everything.
March 7th, 2008 at 6:47 PM
@9 Wake up, you must have been using the buggy WindLight Client which is now outdated and users of the client have been forced to upgrade to the new Test client. Try download the STABLE viewer instead if your not satisfactory continuing your bug hunting on the buggy viewers..
http://secondlife.com/community/downloads.php
March 7th, 2008 at 6:57 PM
@14 ty for trying to be helpful but no…using latest hardware and software…not talking to a beginner here…bottom line is sl needs to ramp it up…seems to not have adults working here who understand what it really really means to deliver reliable products to customers…sorry to say this…but many many months of experience have proven this true over and over and over…do u understand?
March 7th, 2008 at 7:07 PM
DEV-10702: Push-based weapons are now functional (the previous deploy had code that would penalize attempts to push “overly hard” to the point that many weapons displayed zero push - whereas now they are simply capped to the max push allowed)
————
Why undo this? People generally don’t like being orbited.
March 7th, 2008 at 7:08 PM
thanks for updating the servers for the new media texture feature.
March 7th, 2008 at 7:08 PM
[...] Lab just announced that Havok4 simulator code has been updated and that it has been released to both the Beta Preview and the currently more than 500 Main Grid simulators in the Early Adopter [...]
March 7th, 2008 at 7:13 PM
@Warkirby - True, most people don’t like being orbited, but content creators also don’t like having their in-world products broken. And it’s possible that such well-intentioned fixes had adverse effects on more benign scripts that used push.
March 7th, 2008 at 7:28 PM
To all re push weapons and orbiting… Push is limited. Yes, it is possible to build an orbiter, but the classic blitz orbit will no longer work. There are people in-world who are quite fervent for basic push-style weapons for combat, and we did not want to completely disable that category of use. /Sidewinder
March 7th, 2008 at 7:36 PM
@15 Wake up in SF: You asked me to IM you in-world. Could you please tell me your Second Life name so that I can do so? I did a check in our user database and have not found a user “WakeUpInSF” or one starting with the string “WakeUpIn”, so I assume that this is a made up alias for commenting on the blogs.
You claim the current client is “totally unsatisfactory”. Please provide me with the specific version of the viewer you are using (from Help > About Second Life”. You can copy and paste the top line. This wil tell me which specific viewer you are running.
Can you provide specifics of the issues that you are encountering. Please note in advance that I will delete posts that are summary rants about everything being broken. If you would like help with something specific, or for me to check something out, I will. Using this blog post as a method of generalized, non-specific insinuations and essentially “blog griefing” is not something that I will tolerate. Such posts will be deleted.
Please feel free to send me a notecard in-world if you have extensive comments or many specific items (Sidewinder Linden in-world).
Thanks,
SIdewinder
March 7th, 2008 at 7:44 PM
@13 Takuan: Thanks for your consideration. I went back to review that bug report, and noticed that there was some internal discussion in the comments that seemed to be about what proper behavior should be, and how to replicate it. I had thought when I last looked at it that our current fixes might make the issue you’ve reported behave better or even be resolved.
Have you checked the behaviors with the code we deployed tonight to see if things are working any better (and if not, could you check to see if it has been addressed by other work that we have done)? If not, please let me know in-world, and I’d be happy to drop in so that you can demonstrate the problem.
Something to note is that we had thinned the physical representation of the avatar somewhat in previous builds, and went back to make it larger again (part of fixing hugger positioning). I wonder if the slender avatar representation was part of what was making it seem that “kicks” did not work well.
/Sidewinder
March 7th, 2008 at 7:53 PM
Sidewinder, since you’ve mentioned avatar representation… why is it that we can’t have smaller avatars? Why do we have to distort bodies to make tiny avatars? Why can’t pixies and leprechauns be small size? Why the continual arbitrary decisions that limit function on SL?
March 7th, 2008 at 8:00 PM
It’s a pity that push based weapons are being allowed again, but if it stops the insane numbers and your scripts being fried and also your avatar deforming and turning invisible then it’s not so bad I guess.
For curiosity, what is the maximum push height, and why in No Push zones should a device be able to get me there? What Rationale was used to allow the landowners choice to be overridden?
SVC-1261 Still happens, although rarer, still slide backwards slowly… also the somewhat linked SVC-1702 still rarely happens.
SVC-1281 The swimmer still hasn’t changed, impossible to swim linden water without this.
You had my hopes up with SVC-1469 but I see that you still can’t move objects that are over 768M without them falling to the old ceiling height.
Definitely ready for release candidate status!
March 7th, 2008 at 8:01 PM
I have an Nvidia 8800 GT vid card and have been running wonderful until this update………now im lagging horribly and have crashed several times today!! I can no longer log into windlight even after I downloaded this update!!!!!!!!!! what is wrong?
March 7th, 2008 at 8:02 PM
@ Sidewinder “Something to note is that we had thinned the physical representation of the avatar somewhat in previous builds, and went back to make it larger again (part of fixing hugger positioning).”
Eeep. can you please clarify this? As you know I am 4′5″ tall O_o you have scared me.
March 7th, 2008 at 8:25 PM
“We will not practice “release and run” with this simulator, just as you have seen that we do not practice “post and run” on the blog…”
Kudos for such statement Sidewinder, I wish I’d see more of that with the other teams, with a spanking from the love machine if they fail to comply to your standards.
March 7th, 2008 at 8:32 PM
SideWinder, Txs for the fixes.
llTargetOmega has a problem though.
it is not rotating linked prims in a physical object.
llTargetOmega is in a child prim.
Was working earlier today before the release.
March 7th, 2008 at 8:32 PM
Looks like SVC-845 is still broken. My shape-shifting vehicles lose steering after a prim change. Other changes look great. Awesome!
March 7th, 2008 at 8:34 PM
wow great job again guys!
thanks for fixing the banking ground vehicles! i love you andrew (or whoever did it)
and i see you worked on the jumping problems too, but sorry i have to say this, it needs more work. actually i have the feeling i jump even more now. somehow… different. not the extreme jumps like before, but the number of jumps seems to have increased.
and i still can lag sims with my skirt
will try to come to the next office hour, couldnt come this week…
hugs
March 7th, 2008 at 8:41 PM
@23 Blinders Off - another alias that does not match any Second Life user name that I can find… No user names that begin with “Blinders” in the user database…
These questions are completely off topic for this thread. If you would like to talk about those issues I would be happy to in-world, and to do that I will need to know your Second Life name.
The quick answer to “why the continual arbitrary decisions that limit functionality in SL” is that that is a self-answering, biased question that does not take into account the way that software and products are built. There are no software systems that do everything and allow everything. All design is tradeoffs, tradeoffs of ease of use, flexibility, performance and various other factors. Some things that seem like limitations some years down the road are not imagined when systems are built.
When limitations are found, there are choices made as to where to invest resources, and which types of flexibility and new features are the best use of those resources. You may not agree with where the functionality expansions have been done in Second Life, but there have been many, many expansions of functionality over time.
At the moment, projects such as the Havok4 project are focused on reliability and stability - and frankly many folks - I believe yourself included if I recall correctly, have complained that we are adding features rather than focusing on stability. For this project, we are consciously avoiding adding significant features in order to invest resources in removal of crash modes, increased stability and enabling some future project to add new functionality through a more stable platform.
Best regards,
Sidewinder
March 7th, 2008 at 8:46 PM
@24 Sean: SVC-1281 - aka the swimmer problem - was being worked on today and I believe is pretty close to resolution. I hope to have that fix in the next deployed version (not been forgotten!).
Push in no push zones should be significantly reduced or removed in many cases - are you still seeing this as possible with this latest deploy? If so, please contact me off the blog so that we can document the situation and see whether it can be handled without breaking other things.
The build ceiling will be raised to 4096m, *but* you will not be able to access that height until there is a viewer update, since there is a cross-check for this in the viewer code. This update will happen after the Havok4-based simulator is rolled out to the rest of the Second Life Grid in order to avoid confusion and issues with a mixed simulator environment.
@26 - Avatar thickness, not height, is what was adjusted
/Sidewinder
March 7th, 2008 at 8:47 PM
@28 Darek: Hmm… Could you contact me with a repro case for this. It sure seemed to be working… Thanks, Sidewinder
March 7th, 2008 at 8:49 PM
@29 Cheshyr: Could you provide me a specific step-by-step with sample vehicle (or pointer to where I can pick one of that model up) so that I can be sure this gets addressed? Off-blog would be great - thanks, Sidewinder
March 7th, 2008 at 8:50 PM
@30 yuriko: I have to grin about “lagging sims with your skirt”… For those who don’t know, this is no normal skirt… it’s a scripted multi-prim animated piece with hud to control it! We have some other performance optimizations in the queue and I hope that they will help with that case. See you at office hours, Sidewinder
March 7th, 2008 at 8:53 PM
Changes in direction of slow physical motion have become VERY jerky and almost seem to freeze for a split second since the update.
March 7th, 2008 at 9:00 PM
Sidewinder if you still around … take a look here:
http://slurl.com/secondlife/Constanta/212/183/22
…the physics on this H4 sim doesn’t work at all, even after a restart (the restart was because this H4 upgrade), the problem is NOT from this latest upgrade (it happened even before). I’ve left a message to you inworld. I did not open a jira entry since i don’t know what’s the problem.
That’s not my sim, im reporting the problem since an user there uses one of my products (but any other physical object doesn’t work there). Cheers.
March 7th, 2008 at 9:39 PM
Ah, a test of orbitting weapons on Push Safe land shows no or minimal action. This is good.
March 7th, 2008 at 9:46 PM
To Sidewinder, Andrew, Simon, and the rest of the H4 team, keep up the good work! I haven’t been able to attend the last few H4 meetings, but are there any plans to commit the vehicular friction change that Andrew considered implementing? (something about vehicular friction completely overriding a prim’s “native” collision friction)
March 7th, 2008 at 9:50 PM
My jetski (Mocassin 1200 v1.0 by Scout Detritus) is still borked. It’s even worse now. I can barely take a turn in low gear. It just wouldn’t move. Considering that this is one of the most popular jetskis in SL that’s bad. The creator is not available for quite some time now. BTW I can’t swim anymore either. SLWIM is broken. Will I drown now?
March 7th, 2008 at 10:29 PM
@37 Vittorio - that is because collisions are disabled on that region
/Sidewinder
March 7th, 2008 at 10:31 PM
@40 Monlisa:SLWim is being worked on right now - literally was started on in the last day or so - so it’s not fixed but is being looked at. Can you IM me inworld at some point so that I can take a look at that jetski? Thanks, Sidewinder
March 7th, 2008 at 10:42 PM
Well, as what I call an anti-griefer (I know of a griefer attack, I go to the scene and file an AR), I know what the code can do, especially in sandboxes. Perhaps Linden Labs could design an all-new simulator all-together designed solely for sandboxes, that allow, say, a 10-20 meter push? Perhaps a kind of limited-push setting for land, that restricts pushing.
March 7th, 2008 at 10:44 PM
Sidewinder, might I suggest you take a look at SVC-1662 (http://jira.secondlife.com/browse/SVC-1662), I just tested it on my Havok4 Beta Server 1.19.1.81747 sim and the problem described is still an issue.
March 7th, 2008 at 10:53 PM
Hey Sidewinder, thanks for the feedback in SL.
March 8th, 2008 at 12:24 AM
Still happy with this version, after much testing. No need to roll it over to the production simulator server from my point of view. This seems pretty solid for me at least.
March 8th, 2008 at 2:10 AM
Keep up the good work Sidewinder, mate
Many of us appreciate it.
March 8th, 2008 at 2:39 AM
I’m aware this is a bit OT, but could you please fix the Windlight Linux Client in a way, that it logs into the grid again? All it does is to crash on initializing world.
March 8th, 2008 at 2:48 AM
Sidewinder, this is great! Thank you for the quick update!
March 8th, 2008 at 3:07 AM
Well, thanks for two things: first, for the “avatars sliding down the slope”. It was both very amusing but a trifle annoying, but I’m glad you worked it out).
Many many thanks for allowing HTML-on-a-prim on the Havok4-based sims too
That was a wonderful surprise on this “last minute” deployment which did make a lot of us very happy 
(I’m testing it right now on one of the early adopter regions, and it works just fine… and I’m not sliding down the slope into the water when looking at a Web page
)
March 8th, 2008 at 4:33 AM
Great :P. Good progress:). Go on like that!
March 8th, 2008 at 6:01 AM
Darn, I thought the sliding was gone, but it is still there, ok not like before, but it is. Here is the repro: stand on a slope and turn AV 180 deg and then back , do that some times and see - moved , even without AO :(.
March 8th, 2008 at 6:31 AM
Okay I have a question, and it’s a serious one.. And while it may seem sarcastic, it really isn’t.. and I’m very curious about hearing an answer.
I’m watching the implementation of Havoc 4 onto the grid, and what I’m seeing is that yes, you’re tweaking it so that old content still works the way it should. This is great.. on one hand, because it may mean that we’re not all forced to run out and make/buy new stuff.
But there’s something I’m not getting here.. if you tweak Havoc 4 to the point that it functions the same as Havoc 1… what benefits are we actually GETTING out of H4? I’ve seen a lot of broken, as several friends have H4 adopter sims. What I’m not seeing is anything that really seems “better”. and the effort seems to be towards making H4 indistinguishable from H1.
If that’s the case… why aren’t we just remaining with H1?
What’s H4 good for anyways? Why are we bothering with all this if the end result is “basically the same as before”?
March 8th, 2008 at 6:32 AM
@ 32, Sidewinder: “…The build ceiling will be raised to 4096m, …”. I’ve not been involved with Havok design, but I hadn’t heard that before, and would like to put in an opinion.
I think 4096 is far too high. 4k would seem designed to allow people and things to disappear into distance. And, 4k high versus 256m sims, is like building 30-story needle buildings on town house plots in Manhattan (been there, seen that
I assume the advocates of 4k are some subset of landowners who want to disappear ;-/, who 1) want to look outside and see nothing, and/or 2) don’t want their activities to be discernable (sp?) by anyone passing by, or 3) who didn’t know how to space skyboxes to prevent chat leakage
Some advocates might think that being up high will prevent anyone from finding them. All i see here is self interest by such advocates — no group interest (SL population, or SL landowners) is being served.
4k is far more than can be populated by e.g. vertical stacks of rental skyboxes, because of prim limits and simple teleport limits.
I’ve lived on my First Land parcel, on an open platform at 700 meters for the last year, and my main store is elsewhere at 600 meters. As a sim matures, the heights form an ecology and scenery that is fun and interesting to wander through. I think that would be lost.
I also think the resulting willful isolation of some people, would be a overall negative to the SL experience. Increasing the existing opportunities for this or that corner of SL becoming concealed from all outsiders, I think is also negative.
Elbow room above cloud layer, of course is good !
Long vacant echoing open-walled elevator shaft, with pigeon nest at the top, imo no ;-(
You could try a survey perhaps, asking Premium accounts or landowners older than six months, or maybe all accounts older than one year?
Restricting 4k to private islands, would also work imo.
March 8th, 2008 at 6:45 AM
@44 ZATZAi: Thanks we’ll look at 1662. That was in the LSL stack so I’d missed it on review but we’ve got it now. /Sidewinder
March 8th, 2008 at 7:01 AM
@53 Winter: The quick answer is: Stability - and the architectural shifts needed to enable the use of new physics features in the future as appropriate. This is, and has not been, a features project.
From the start of the beta process I’ve been trying to be very clear about this… This is not a “features release”. The goals are simple:
1) Reduce incidental crash rates
2) Reduce content loss caused by rollbacks from incidental crashes
3) Remove targets for purposeful crash vectors (griefer-induced crashes)
4) Move to a more current version of the Havok physics engine and reduce the effort to stay current
5) Enable (but not yet implement) the ability to use the Havok4+ new features in Second Life by moving to a more current version
The Early Adopters have so far found that the incidental crash rate is significantly reduced from the levels seen on Havok1-based simulators. We are working to further reduce the possible crash vectors before release. Second Life product creators have been consistent in saying that they are no longer seeing incidental simulator crashes based on vehicles, toys or animals doing something that the physics engine or simulator “didn’t like”, which dramatically reduces the need for rollbacks and reduces content loss.
Two more side benefits related to stability are that with the simulator now this “clean” from a crash vector perspective, identifying and resolving new crash modes is getting more straightforward.
There area small number of remaining purposeful crash strategies - in other words crashes that do not happen by accident - they are triggered by someone trying to crash a simulator (and I know that you griefers watching the project know about them too, and have been witholding proof of concepts during the beta… sorry guys we already know the techniques you have in hand). The remaining purposeful simulator crash vectors follow a similar pattern, and frankly “leave fingerprints” on the simulator that make it pretty clear who perpetrated the crash. In addition, in most cases, these crashes do not require a rollback to resolve. We will be closing those loopholes as well, but have focused on incidental crashes first, since they affect the highest number of residents on a regular basis.
In the process, the linkage rules have been changed to be a spherical bounding model, so that prims linked at the boundary of the rules may be rezzed in any orientation in world without “falling off the edges”, whereas the current simulator uses a cubic bounding volume that is orientation-sensitive at the boundaries.
Hope this helps,
Sidewinder
March 8th, 2008 at 7:04 AM
Can’t wait to see Havoc4 rolled out. Good luck
March 8th, 2008 at 7:05 AM
Can’t wait to see Havoc4 rolled out. Good luck
March 8th, 2008 at 7:10 AM
@52 Sascha: Thanks for that repro re sliding. It seems to me that we’ve resolved the fundamental issue, and that this remaining case should not block integrating the simulator with release. As mentioned before, that does not mean that it would not get fixed, but that it might end up as a normal maintenance fix rather than something that prevents the rest of Second Life from getting the benefits of this update. (If you were making the call on this, and you’d hold the release for this fix, please let me know why - I’m happy to be convinced if I’m missing the wide scale impact that this would create.)
Regards,
/Sidewinder
March 8th, 2008 at 7:23 AM
Thanks for this Havok 4 update. Happy to play with the websites on prims, been wanting that! And it is running great so far! As well with the new RC Viewer I have had no problems running in this config:
Second Life 1.19.1 (0) Mar 5 2008 17:51:19 (Second Life Release Candidate)
You are at 263211.6, 299030.6, 22.8 in Inochi Island located at sim5621.agni.lindenlab.com (8.2.34.177:13005)
Havok4 Beta Server 1.19.1.81747
CPU: Intel Core 2 Series Processor (2399 MHz)
Memory: 3071 MB
OS Version: Microsoft Windows Vista (Build 6000)
Graphics Card Vendor: NVIDIA Corporation
Graphics Card: GeForce 8400 GS/PCI/SSE2
March 8th, 2008 at 7:31 AM
Re: Wake Up In SF……..
If you cannot use your SL name you are not worthy of a response from the Havok4 group. Sidewinder and his team have behaved in an exemplary manner throughout this project and have displayed a genuine willingness to respond in an extremely constructive manner to resolve genuine concerns and to correct problems. Sidewinder even had the good manners to respond to you, which you certainly did not deserve. Either use your real name or butt out !
Well done Sidewinder…..you and the team are true professionals.
March 8th, 2008 at 7:36 AM
@54 Dwayanu: I understand your concerns. At the same time, if there are residents who would like that space for building, for whatever reason, or even want seclusion as part of their Second Life, why should we prevent them from having a world that accommodates that life style? This seems akin to requiring participation in group events in order to mandate community involvement. It seems to demand a specific lifestyle in the form of property that seems antithetical to the goals of allowing each Second Life resident to style their part of the world to their desires.
I may be missing the point, but I see this as a matter of inclusiveness and respecting the desires of a widely varying and global community to express themselves and style their property and the air above it in whatever fashion they would like to (within the TOS…).
Also please note that on larger parcels you can place much larger platforms. A large platform is as visible at 4096m as a moderate to small platform is at 768m. It is not completely clear to that “wanting to disappear” is the only reason for high altitude placement. It could also be that with a large platform you don’t want it to have too much visual sky obstruction while still being visually present.
Something thing I’ve learned about Second Life as a platform is that people will find ingenious, unique and amazing things to do with each capability of the platform. I suspect that we’ll see all sorts of “cool stuff” done with an expanded build ceiling that we cannot imagine today. As in the real world, there may be upsides and downsides.
At the same time, I do see your points, and will continue the discussion on this internally before release.
/Sidewinder
P.S. The current build ceiling is at 768m, not 256m, which for small platforms is pretty close to invisible already. This is roughly a 5x height expansion. As part of this work, we’ve allowed for faster assist speeds (256m/S) - in particular because builders who have been involved with the beta process wanted to be able to create short-range teleporters that would cover these distances in a reasonable amount of time. This was the topic of many discussions over the last few months in the project office hours, which are still continuing.
March 8th, 2008 at 7:46 AM
@22 Sidewinder - Yeah, my explanation on the JIRA issue wasn’t very good, was it? I will head in-world today and test to see if the other issues that were fixed resolved this one as well, and if not I’ll do my best to come up with a repro script that can concretely demonstrate the issue.
March 8th, 2008 at 7:55 AM
I’m liking the updates so far. Not sliding into the water while talking to someone next to the lake in Serenity Woods is nice, and that temporary prims no longer count towards the prim limits is awesome. I believe I’ve found a small issue with binding a mass (512) of objects to the physics simulation at the same time however. Apparently when I converted my “cube of cubes” (an 8×8x8 cube made up of cubes, 512 in total) to temp/physical some of them reverted back to non-physical/temp (they rotated slightly out of place, but then froze in mid air and didn’t disappear along with the rest), while others seemed to spin off into the abyss, slowly, like the sim was outer space. Like I said, small issue, I don’t really know of a practical application for adding that many objects to the physics simulation at the same time, aside from relieving stress by toppling towers.
I do have some questions however regarding the linkage rules and prims , if you could contact me in world or something. I promise not to take up too much of your time. >)
March 8th, 2008 at 8:04 AM
@22 Sidewinder - The behavior is definitely different now, though not quite correct. Previously a kick against a standing opponent in Havok-4 would do nothing at all, now it does push them just a bit in the air vertically, but with no horizontal movement. I sent you an object that can repro that with just a click, and the script inside is simplified to (hopefully) make it clearer how it was intended to work in HAVOK-1.
The other cases mentioned in that JIRA issue in the comments may also be changed, but I’ve not yet had the chance to find out.
March 8th, 2008 at 8:08 AM
I guess no cure for the horrible lag, freezin and crashing in the Rome/sparta sim , eh??? Talked to several after we raided last night and many of them are having the same problems which werent there before………yikes!! Makes me get killed !!! ;o) Hate when that happens! please offer any assistance you have…any help is GOOD help!
March 8th, 2008 at 9:11 AM
@66 Candi: Which regions are you talking about in your post? There are a few that have Rome and Sparta in the name, and some are Havok4 and some are not… In particular it is also important to separate out viewer issues from server issues. If the whole simulator did not crash, in other words all avatars were not thrown off the region at the same time, then this is a viewer, network or data center issue. If you will provide more specifics it might be possible to find out what was going on, and to determine whether this had anything to do with the Havok4 simulator or was part of a general infrastructure or data center problem.
Thanks,
Sidewinder
March 8th, 2008 at 9:19 AM
@64 Feynt: You may be seeing some of the new code that throttles the physics engine when it begins to get overloaded. It’s one of the mechanisms to prevent the simulator from getting bogged down and freezing or crashing.
We’ve found two general things slow down the region in physics - collisions and changing shapes, so we’ve added code to try and control this to keep the region alive.
A few things may happen if the region gets too slow:
a) It will lower the ‘level of detail’ used to model objects. It will start treating some objects as simple boxes rather than the full-detailed mesh shape
b) If it gets even slower, some physical objects will be temporarily frozen in place.
c) If it gets really slow (over 2 seconds per frame) it will start ignoring some collisions compeletly.
All this is to lower how much calculations it needs to figure out the physics in world. As the region recovers to a higher fps, the original settings for objects are restored and hopefully everything returns to normal.
I’m guessing you’re seeing some of this effect - it sounds like some of your boxes are being frozen for a bit. I’ve actually used a very similar situation (a grid of boxes, and setting them physical) when testing the code.
If you see behavior that can’t be explained by this, definitely let us know.
March 8th, 2008 at 9:53 AM
Just wondering if there will be any more regions added to the early adopter program? Apologies if this was covered already, I’ve read through and haven’t seen a reply to my question up in #5. Thanks again!
March 8th, 2008 at 10:30 AM
We are a Havok4 test group, I guess you would call it………if I can log in I will post the info here………thank you for answering me.
;o))
March 8th, 2008 at 10:31 AM
Keep getting DNS log in issues, and Im on the internet right now!!! I will keep trying .
March 8th, 2008 at 10:33 AM
That is a smart idea, Simon, however I’ve got a few comments/questions on those steps:
a) Some shapes would be better resolved as spheres (actual spheres, for instance, or a torus), and as far as I’ve learned spherical collision detection is easier to calculate than box collisions. Since you only really need one check (llVecDist(pointA, pointB) D
My questions weren’t about these though, but I don’t really want to ask in a public forum because it might illicit poor feedback.
March 8th, 2008 at 10:36 AM
heres the info ……………..THE SPARTAN EMPIRE, 68,135,23
We were having a sparring match in the grass off and on all day yesterday , from what i could gather from the others was horrible lag, so much so, you couldnt tell the direction you were going! Intermittent freezing and then crashes……..;O( Sorry for the probs, I hope its not too big of a headache… ;o)
March 8th, 2008 at 10:49 AM
Bah, it ate my comment. WordPress bad. x.x
b. It seems like it doesn’t hand the object back to the physics simulation when lag subsides, about a dozen of the prims remained while about 500 of them settled or careened off into the sim with no gravity to stop them. That dozen, give or take, just stayed floating in midair without disappearing, which leads me to believe they were completely disabled, not just removed from the physics simulation.
and c. I didn’t notice any prims ignoring collision, most of them settled on the floor, that dozen or so got disabled midair, and…. Oh I don’t know, two dozen went slowly sailing off into the abyss like gravity was disabled on them. I believe perhaps the engine is trying to cancel out all forces on those prims, but failed to take into account current forces, OR it’s getting forces applied to them due to interpenetration, which you’re not stopping and seem to be forgetting to add gravity back to those prims.
March 8th, 2008 at 12:16 PM
@59: I can still reproduce the sliding, but it seems to be depending on who tries it. A friend and I tested it and while I slipped down the slope, she was standing firm. A steeper slope made us both slip, but after changing avatars she stood firm again.
Another issue I found: Walking with space bar held down is not possible on Havok 4:
https://jira.secondlife.com/browse/SVC-1786
Zi!
March 8th, 2008 at 12:28 PM
@ Sidewinder Linden: Sidewinder, to my recollection, Lindens aren’t all that easy to get hold of in world, are they? (Another major customer complaint).
If you people want folks to use their in-world names here, rather than chiding others for using alts here… maybe you should allow people to sign in only with their account names. Pardon me, but DUH. I use the Blinders Off tag here for a good reason: to constantly remind people not to swallow every word that comes off the keyboard of Linden Lab. That is my exercise of freedom of speech and is prompted by past propaganda from Linden Lab offices (no insult intended… plain fact. You DON’T want to ask me for examples. This blog doesn’t have enough room.)
What I mean by arbitrary decisions (again without taking up miles of space to listing examples), the 10m prim limit (accompanied by totally tech-bogus explanations as to why that is necessary), the current avatar size limits, the flight limit (we can build up to 700m or so but we can’t fly that high? And so almost everyone owns a flight device so they CAN flight that high? Come on).
Shall I go on and on? Sorry Sidewinder, but between our two posts, I think mine was far the more legitimate and honest. Of course, you can delete this post, but it won’t be because I’m unnecessarily demeaning, rude or flaming in my post. Everything I’ve said here is factual and right on the button. And while I’d like to deal with LL in a professional, friendly manner, it seems after several years of experience, LL only tends to listen when it’s hit over the head with a board. Sorry, that’s just my personal opinion (and that of many others). Why do I say that?
How many people have NICELY asked for Linden Lab to fix group chat problems. And for how many months have they been nicely asking? (simple answer: more than a year)
How many people have asked LL to fix the llTargetOmega glitch in the Linden Scripting Language? And for how many months and months have they been asking?
So please, don’t try to make it sound like all we have to do is contact a Linden in-world if we have issues. Frankly guy, that’s pure nonsense. (and yes, some LL employees do have “office hours”… what an hour or two a week, during business hours when most people are at work…)
Sorry Sidewinder, honestly no offense intended, but it doesn’t work that way.
March 8th, 2008 at 12:32 PM
BTW Sidewinder, as a side… you folks are expanding ceiling from 768 to appx 3500m? GOOD JOB. (See… I’m capable of a pat on the backwhen warranted).
Will we have flight ability to 3500m as well? Hoping so.
March 8th, 2008 at 12:38 PM
@ 54 re 4096m ceiling (yup 4096 not 3500m like I’d heard. Woot!)
I appreciate your opinion and agree somewhat with your feelings, but in truth, I would find people isolating themselves at 4000m much preferable to the constant “security barriers” we slam in to while flying through mainland.
There are many advantages to high-sky building. Groups may wish private discussion chambers that they can secure. Some people may wish high-sky builds in which they can have some REAL privacy. With 4096m, a sim owner can grant land renters certain sky levels along with security device rights… and 4096m is enough to grant them 100m all the way around without causing sky space problems.
I think this is a good move by LL. Perhaps it will help remove some of the garbage and endless security grids from the surface.
March 8th, 2008 at 12:47 PM
Sidwinder, forgive mulitple posts, but while you folks are bringing in Havok 4, it sure would be nice to be able to TP anywhere on a sim without having to use a LM. The current limit of (what, about 296m? not sure) may have a tech reason behind it I’m not aware of, but if that could be overcome, that would be spiffy.
March 8th, 2008 at 1:27 PM
SVC-1488: Prim.Docker automatic prim alignment no longer leaves gaps
I did some general testing and it’s behavior is perfect.
I’ll show up with my first born during the next in-world office meeting.
March 8th, 2008 at 2:31 PM
@69 Atashi: I did see your post about Early Adopter additions, and have been checking on it. The Concierge team is handling these on a time-available basis and is getting to them as fast as they can. I suggest that you contact the concierge team directly if you have an urgent need. /Sidewinder
March 8th, 2008 at 3:06 PM
@79 Blinders Off: I don’t fully understand what you’re looking for in terms of TP to anywhere on a region without a landmark. You can teleport to anywhere on any region by typing coordinates into the map panel… so unless I’m missing the point you can already do what you are asking for… maybe I am missing what you meant? /Sidewinder
March 8th, 2008 at 3:50 PM
I’m really looking forward to grid-wide implementation of Havok4 physics, and I would really like to be able to participate in the bug-testing and feedback. Which is why I volunteered to become a first adopter. Unfortunately, Havok4 physics is active indiscriminately, or at least it’s bugs are, on objects whether or not they have their physics on. This makes it unfeasible to monitor objects to make such reports.
C’mon, I’ve got a support ticket requesting a rollback to the start of the opt-in period, and it’s been gathering dust over the last week. Do I need to open another to get attention?
March 8th, 2008 at 4:03 PM
@79 Bilnders Off: By the way - the reason I had hoped to find out your Second Life name is that I suspect that if we can talk inworld, vs “one comment at a time on a blog” I might better understand the issues and you may find that I share more of your goals than you suspect, and as a result we can both be more effective at achieving the goals than in this forum in this manner. Regards, Sidewinder
March 8th, 2008 at 5:03 PM
@59 Hehe yes is ok, i don’t think that so many people will go to beaches with slopes from terrain and try to hug or kiss people and slide away. i will just think that i am standing on slippery ground and wait for a maintenance fix :P.
March 8th, 2008 at 5:19 PM
#83 Prodigal: I just found the jira you entered, SVC-1384.
My apologies for missing this one - it was marked normal priority as opposed to some higher priority and for some reason I missed it. We had found a problem with objects of certain geometries or placements being mishandled at startup with Havok4, and believe that we have resolved that problem.
For what it’s worth,It turns out that our Havok4 startup routines were being quite strict about validating object integrity, and not tolerating some construction problems that earlier release simulators had allowed. We loosened our startup checking to be sure that legacy content would load correctly, and left the strict checking in place for new construction.
Lets get in touch inworld so that we can set up a time to work this out? I don’t think that this one is a simple concierge simstate rollback, and apologize that it was not flagged as a develepment team issue from the start.
I would like to pull a simstate from before the conversion and be sure that your region will load correctly with Havok4 on the Beta Preview. If it does not, we will resolve this issue - prim content loss is for us a serious issue that we work at very high priority and had thought that there were no remaining issues.
Thanks for bringing this up, and I will make sure this gets addressed.
Regards,
Sidewinder
P.S. I’d also like to talk more about “physics is active indiscriminately” as I am not sure what you are talking about with this comment, and would like to be sure that it is either already addressed in the current build or that the issue gets clarified so that we know where to look for a solution. Thanks again.
March 8th, 2008 at 5:24 PM
The slide is not fixed for small, (probably low mass avatars.) In my normal 4′5″ self I will slide backwards, on even a minor slope ( http://img80.imageshack.us/img80/9135/snapshot001va8.jpg ). No special steps required, just stand there.
When I put on my 6′ shape I don’t slide.
In answer to the question posed above… and knowing the way the average SLer likes to complain I would still roll it out based on the benefit of blitzing being killed off. Slide down a hill if you are a small avatar or be orbited by people who hate Kids is an easy one.
March 8th, 2008 at 5:56 PM
are Teleporting drop pointed fixed? its a bad when you telport to a location where you KNOW where the drop point is then you landup in a different location.
March 8th, 2008 at 6:35 PM
Usagi, I have never seen that problem unless the land owner has forced a landing point. Do you have a Jira reference for your issue? Does it only happen on Havok4 sims?
March 8th, 2008 at 7:04 PM
@87 Sean: Thanks for that catch… the fix must work down to a certain avatar mass and then not below that.. I’ll re-open the internal bug on this one. That would hopefully also resolve Sascha’s finding as well. /Sidewinder