Joints?

Eep, this post by Andrew Linden should hopefully explain things to your satisfaction. It has already been linked to in this thread at least once so I thought maybe you had already seen it. Anyway, here it is:
/108/39/86645/1.html#post881905/108/39/86645/1.html#post881905
The third paragraph is relevant to your question.

Persig

Two have posted to the thread - I had exactly the same query and Andrews posts were very informative.

It's the programming equivalent of putting something important from your in-tray through a shredder.

There is a big long list of 'known bugs and issues' that need fixing. By effectively removing the problem, then it's no longer a 'bug or issue' that needs fixing. Resulting in a 'the game works wonderfully because there's hardly anything that we know doesn't work how we intended it'.

It's much easier to just kill off and forget about a problem than to actually try to fix it. Who cares if it upsets a few players and they quit, there's plenty more people signing up to play.

Lewis

Thanks; missed that. While that DOES explain things, it still makes me wonder if joints will EVERY be reimplemented after Havok 3's implementation. How long will Havok 2+'s 4TH(!) attempt take? I'd hate to be without joint creation for at least 6 months!

I will openly admit that after a few years here and seeing Havok 2 (thanks for the correction) become vaporware I still have very little doubt that IF it finally gets implemented joints will certainly not be far behind. Actual working joints would be too damn useful for them not to be high on the priority list. My guess is implementation in Havok 2 or 3 is going to be orders of magnitude easier than it ever was bfore when they had to support the original joint system in parallel. Again, I'm admitting to being a wide-eyed optimist on this one but I think they know how much content creators would like working joints...

Persig

Eep's question is "How does removing joints NOW help the Havok-2+ implementation?"

Removing joints NOW doesn't help the Havok-2+ transition (which is still a couple of months away) so much as it hopefully reduces the amount of jointed content that would suddenly break LATER when the Havok-2+ transition happens (assuming we actually do break joints in Havok-2+). That is, we want to phase joints out of the picture instead of cutting them out suddenly.

Here's our plan-A:

Step-1 is to tell the SL Residents what we're thinking and why (that is where we are now).

Step-2 would be to disable the UI that allows joints to be built (joints rezzed from inventory would still work).

Step-3 would be to remove support for creation of new joints (this would be the point at which they actually break, and this would happen very close to the Havok-2 transition).

Step-4 would be to re-implement joints better than they were before.

Plan-B would be to bite the bullet and support joints through the Havok-2+ transition, and perhaps eventually clean them up (without breaking them) so that we can add better functionality. We would attempt Plan-B if the cost of removing joints appeared to outweigh the cost of supporting them.

Since we are currently in Step-1, let me present the reasons why I'm in favor of Plan-A:

Eggy asked earlier why we didn't just scan the world and count the number of joints being used and thereby measure whether anyone was using joints. We did a while ago. As I recall, less than 10 active joints were found in-world between 1500 regions, about 10K land owners, and 40K active Residents. This is a testament to how useless most people in SL find joints, and one reason why Plan-A might be feasible -- it was very likely that few Residents would complain. Nevertheless, we are empbarking on Plan-A Step-1.

The current implementation of joints in SL makes it difficult to add new features or to re-organize code because the joint code is scattered about in places where joints should not need to be considered. The reason for this is that joints were first done as a quick hack back when SL did not have many of the features that it has today or -- before groups, sitting on objects, and inventory. As the features were added they had to support joints and special-case code was added to handle the various cases where joints clashed with other features: What happens when you sit on a joint? What happens when you're sitting on a joint that crosses a region boundary? What happens when someone takes the joint you're sitting on into inventory?

Supporting joints over the years has been a defensive campaign. So far we have tried to make them work most of the time, but have not enhanced them at all. Wouldn't it be nice to be able to edit joints with ease, translate and rotate the pivot points just like you might move linked prims? Wouldn't it be cool if you could make long chains of joints, or if you had a springy joint? Would you like to be able to create and destroy joints using LSL calls?

In order to support any of that we would have to sprinkle lots of special-case code throughout the codebase. Meanwhile, some corner cases would undoubtedly be missed and those would show up as bugs. In other words, it could be done, but it would be a lot of work, it would be lengthy, and frustrating.

In the past, when we've started working on Havok-2 I've always left the joint stuff to the last because I fear it. Havok-2+ handles joints in a slightly different way than Havok-1, which means that we would have to find and change most of that special-case code contributing a week or two to the whole effort. It could be done, but it would be a lot of work.

It would be possible to rip joints out and re-implement them after Havok-2+. If we could do it fast enough we could include them with the Havok-2+ transition and continue joint support unbroken. However, it isn't clear how fast it could be done, and it would be a bummer to hold the whole project up while it waits for joints.

So basically we're interested in taking the fast easy path at the expense of anyone who happens to be using joints. However, it is a significantly easier path, we hope there aren't very many people who would be adversely affected, and we promise that the lack of joints will be temporary.

Edit for syntax error.

You sold me.

Andrew Linden should get an "Employee of the Month" award for crafting such a clear and concise explanation to the residents.
I see that some of my speculation was right on and some was not. No mentions of the shortcomings of Havok itself means either: a) Andrew studiously avoided passing the buck on that one or b) Havok is not really responsible for wonky joints. Either way, this actually gives me hope for the future implementation.
Your wide-eyed optimist,

Persig

OK, I feel better about joints being removed now (thank you for the detailed, outlined plans!) but I DO hope they will be reimplemented ASAP. However, I'm still unclear about:

How would removing joint CREATION not affect rezzed joints? Or does "creation" mean "rez"?

How long do you believe this 4th Havok 2+ implementation will take?

Out of curiosity, how long ago was this scan performed?

As Andrew says, it's more like "joints were a quick hack, and ideally we can use this opportunity to reimplement them right."

That must have been quite a while ago. There's 6 active joints just in the wind-chime in Dwellget, there were another half a dozen joints in the playground in Sables dAlliez until recently, and I know I've seen more than ten swing-sets and wind-chimes in other places.
I'll buy that.

If you can get it there in time, will existing objects be able to come back out of inventory and have a chance of working? Or will joints have to be re-made using the new joints?

If killing joints speeds Havok 2+ integration, joints should die. Hand me the gun. I'll kill em myself.

Andrew:
Thank you for the informative post - as I once told you inworld, you are widely known and loved for those. Frankly, with all the people Robin invites into the community team, you do a better job of keeping in touch with us than all of them put together... because you actually know what you're talking about. It's great to be social but that's not worth a damn thing if you're not also a techie.
It's great to see the actual numbers and it's very comforting to know that LL increasingly keeps track of and uses objective metrics in their decision-making process. While I appreciate LL's willingness to keep us in the loop, as someone once said, the worst way of finding out what a person wants or needs is to ask them directly

NO!

I use joints (hinges), I use them successfully (please, come and try to send one of my revolving/swinging jointed doors to mars -- do a Find Places for "Boso", I welcome you to crash a plane into my hinged physical non-phantom door). Watch the sim stats and see the massive hits my joints create when working/abused (magnifying glass might be required).

Admittingly there aren't many "good" jointed objects in the world. I've only installed about 5 revolving doors. But I'd be upset if joints were removed.

Joints suck, yes. But they CAN work with scripted routines to check for erratic behavior.

I applaud any work towards Havok2/3, but how about removing the old joints when you implement the NEW joints?

Maybe joints need a license to use until you implement better code (havok2/3) -- just don't remove them! Again, please come look at what can be done with the "buggy" joints -- they do work, and work well (I've yet to shoot one off to mars or beyond once installed with scripted behavior checks).

Bosozoku Kato

(posted the above after reading the first 10 or so posts -- then read it all, in particular Andrew Linden's informative input).

Okie, I agree, do what's best to speed along Havok2/3 (although I long ago stopped holding my breath). As long as joints return. Really I'd rather they stay "buggy", so those few of us that actually figured out methods to reliably use them have a unique toy to show off -- but for the good of general (ab)use then I'd have to agree with what Andrew has posted. Sounds like a benefit to the development side.

I'll be upset, sure, when my doors stop working or I want to add a new one, but I can handle waiting for H2|3.

Bos

Yeah, but that doesn't really excuse broken joints; it does makes me amazed you were able to use them so well!

That said, I DID stop by and break it with the combination of Max Case's "AngelCubeDeathHand" physical primsweeper and a gravity gun. the door went into the collusion overload mode, then it disappeared. Dunno if it shot off to Mars, mind... I'm assuming that's part of your scripted safeguards, and they're well-done indeed. I could have probably gotten around it if I was prepared to lag out the sim with scripts to slow down the scheduler, but that's kind of extreme. But you asked people to mess with the door itself, so I did. EDIT: Apparently, the door was still there, it just disappeared locally and reappeared - still in overload - when I left and came back.

I'd much prefer we get a new Havok with new joints that more than a tiny percentage of SL can use, as impressive as your work is!

Aliasi, thanks for swinging by and abusing it -- and for the kind words. Yes it disappears, this is a client-side bug since v1.7 (all my hinged objects draw extremely poorly, post v1.7, and will totally and/or partially disappear -- VERY annoying (and I've reported it long ago)).

You can sorta break it, as you found out. But all you do is cause its scripted "errant behavior" checks to fire -- which, thus far (over a year I think of operation), has kept the door from flying off to mars (actually breaking).

I do agree, I'd LOVE better hinges and eagerly await them -- if and when they do finally get in SL. My point is that hinges do work -- but definitely are not bug free nor easy to work with (in fact they're extremely difficult). I'd just be bummed if they turn off hinges and it's years and years still waiting for "new hinges" in H2/3, which is already years and years in waiting.

It's not my intention to be overly dramatic, but the 100% truth is that if hinges go away, I will do something rash.

In SL, I am a builder of useless machines, and without joints (hinges at least), I have no purpose in Second Life and will have no choice but to END my Second Life.

I don't know how I'll do it. Jumping off a building won't work, and neither will anything you find in Rausch. I suppose I'll do it by contacting my credit card company and telling them not pull the plug.

Yes that's it.

Seriously, despite their many flaws, everything I have to live for in SL is jointed, and without them, I'll be checking out.

Didn't Philip just say that that Havok 2/3 isn't a priority? Guess we'll have to wait a while for those joints.

The first time I heard about SL was a podcast of one of the devs talking about how creative the users were. I distinctly remember him talking about innovative protests they would stage. In one case, he said a group was protesting in favor of upgrading to Havoc 2; something he went on to say is symbolic of a thing that will never happen, as in:

User: "When will X happen?"

Dev: "That's included with Havoc 2...ha ha ha ha."

So now I'm expected to buy it when the very thing upon which my continued status as SLer hinges is, in fact, the arrival of Havoc 2?

I'll consider keeping my account on the one condition that you assign a specific date to the expected implementation. Otherwise, I'm gonzo.

integer Havok2Time = llGetDate() + 6.months;

llSay(0, "Havok2 will be released in" + Havok2Time + ".";

You will be missed.

owch, Phil. I change my vote: BRING BACK JOINTS!

Oh, it's too late? dang...

Is jointing even necessary with Havok 2/3?

AFAIK, a joint is simply a type of relative-motion constraint that locks the physical rotation of one object relative to another object. It is a form of cheating when you don't want to spend the computations required to calculate the forces of a prim-based pin rotating inside a prim-based hole.

A prim-based hinge is far more computationally intensive than a hinge joint, since the prim-based hinge involves friction calculations between contacting rotating surfaces, as well as force-vectoring of the contacting surface polygons between the pin and the hole.

If the newer Havok engines are sufficiently powerful, then they could handle real-world pin-in-hole physics calculations directly without using a computational-shortcut like the current SL hinge joints.


SL with Havok 1 can do real physics with no jointing, such as chain links, but the engine usually freaks out because the calculation load is not optimized, and the objects can get out of sync. The further out of sync the objects get, the more they tend to overlap, which drives the engine "Deep Think" problem Andrew has mentioned before that causes the objects to fall apart, go non-physical, and/or self-destruct.

Havok 2/3 apparantly don't have the Deep Think problems of Havok 1, and both appear to be capable of resolving overlapping prims, so doing real-world tight-precision hinges may be possible without any motion/rotation-constraining shortcuts like hinge joints.

And so, I look forward to making real-world ball-in-socket point-to-point joints with Havok 2/3.

-Scalar