|
History
| Log In
|
|
|
|
|
|
[
Permlink
| « Hide
]
RobbyRacoon Olmstead - 21/Feb/08 08:49 PM
Change priority to Major because the problem indicates a major impairment of function, and there is no possible workaround in a het-grid : There is no way to adjust the push power based on current simulator physics engine, as there is no way to discover via script which kind of physics engine is enabled.
I've noticed some other odd effects with the kicks. Sometimes repeated kicks will juggle an opponent a foot above the ground and if your opponent is jumping into the air or landing a jump the kick will send them flying. The flying backward happened under the old engine too but it seems to be worse now.
Voted and watched. I've seen the same weirdness Malachi mentioned above. Also, if two fighters kick each other at exactly the same time, it can cause both avatars to fly backwards 5m or 10m and up about 3m.
So it might appear that push has some effect when the opponent is in the air? It can easily be demonstrated that it has no effect when they are not.
I dunno if it has to do with kicking an opponent in the air, kicking a moving opponent, or two opponents kicking each other at the same time which makes the kicks push *against* each other at the same time.
I forget who I was fighting in Samurai Edo...maybe Malachi...but the repro was we both ran at each other and kicked each other at the same time while running. We *both* flew back and up. Did it like 3 or 4 times in a row and we flew back and up each time. We weren't trying to test it just so happened that we kicked each other a lot while fighting. LOL Okay, so it's not quite as simple perhaps as I made it appear when I filed the bug.
But... We can agree that it doesn't work correctly in Havok 4, am I right? I really want the Lindens to look into this issue and not blow it off, and want to make sure that they have a good starting point to begin figuring out the causes and hopefully the solution. It is extremely easy to demonstrate with 100% consistency that an avatar that is standing and unmoving when kicked will not be affected the same way in a Havok 4 sim as they will in a Havok 1 sim. The other effects mentioned here are just as important, but are not as consistently and easily reproduced, and it's my hope that researching that case first will lead to a better understanding of the causes of the other cases. My thought was that knowing about the other kick-push effects might help the lindens zero in on the cause of the problem. I've seen it a lot so I'm thinking the repro might not be too hard. If one of you would like to, drop me an IM and I'm sure we can create a simple and reliable repro for the other kick effects. It may be a separate issue but I don't know. I echo robby that it's definitely different than Havok 1 and definitely very problematic, not just because of the lack of push in the kick which is itself a big blow to c:si, but because when kicks juggle the other person is practically trapped.
I think for some reason the push is only effective when an avatar is in the air. If you manage to kick an avatar and land a direct push bullet right after they've jumped. Sometimes it will send you flying back almost 30-40m. For the most part though I can deal with that. But the basic push not working I can't. It really really makes things hard, and I think puts a major nail in the side of the C:SI system. The current state of things only being able to kick someone up, and bounce, or juggled as its been named is quite detrimental I think.
That sounds like a pretty good description, and I agree that this is an extremely serious issue for C:SI, which has many thousands of users, but surely it's symptomatic of something that will affect potentially many many more people.
I'm really hoping that this issue is taken seriously, instead of being punted on, since a heterogeneous grid makes it nearly impossible to create a clean workaround. Pushing in general has been severely limited in H4, mostly in an effort to prevent orbitting.
Scripted attachments pushing the same avatar wearing the attachment (like flight assist) still seems to work, although in H4 it is still limited. As far as I can tell, all other uses of pushing (other than pushing yourself with an attachment) are dramatically different in H4. Even pushing yourself with your own object is so limited, that it will break lots of existing content. Pushing other objects and avatars is also drastically limited in Havok 4, as this post has noted. This is just one of many legitimate uses of pushing that will break. I vote for a fix. All this -does- need, actually, is an increase in push force in the scripted object. Havok severely lessens the power of a push if the target is more than 10m away, but at closer ranges, H4 will push fine if appropriate force is used (tho it needs more force than in H1 methinks.).
If you really want the needed values adjusted server-side, I suggest visiting the official Havok 4 office hours in-world, and pointing Andrew and Sidewinder Linden at this JIRA. They're -very- helpful, and usually immeditely comment on the probability of fixes, and allow a great deal of user input. They've also been actively working with concerned product creators, and you will find that in the changelists shown on the official SL blog, that they specifically mention fixes for products shown as non-working to them. With the particular issue I had in mind while reporting this, the target is *always* within two meters, so H4 does *not* in fact "push fine" at less than 10 meters.
Also, an increase in force *might* help, though I'm not convinced. Another of the side effects of this problem is that the target avatar is no longer being pushed horizontally as well as vertically (if they are still on the ground when the push is applied). Code that used to push the target up about 1 meter and back about 3 meters now pushes the target vertically about 0.25 meters and horizontally not at all. I'll definitely try ramping up the force, though I'm not hopeful that it will fix the problem. I've already given the Havok4 team a super-simple scripted object that can duplicate the issue with 100% consistency, and all they have to do is touch it. Not hard to use to compare H1 and H4 behavior. I should not need to hunt down additional Lindens hoping to finally find one to take this issue seriously. Please give me a copy of the super simple object you mention. I'm having trouble tracking it down.
Speaking to the issue:
Push behavior has changed significantly since this bug was first filed. Not saying there aren't still issues. Have you tested these on the build currently on the Beta grid? Kelly, I just passed you a copy of the object. I don't have time at the moment to test on the Beta grid, but Havok4 early adopter regions still exhibit the problem.
The object should be easy to identify, it has "JIRA I think that yes, you will have to adjust the push values and also that there is a problem (bug) with friction here that is stopping the horizontal portion of the push.
A work around for this might be to double-tap: apply a push up before applying the desired push. Okey dokey, well... Would be nice to get that bug with friction fixed, I'm sure there are other side effects which are so far unreported that would benefit also.
As for double-tap, that is a hack workaround. I'll use it if I have to, but it's a hack :) Adjusting the push values I think the C:SI devs are already expecting by this point, but there's no point in doing that *at all* until there's only a single flavor of physics running again. One of the biggest headaches the issue has caused us is that there's no way to adjust behavior in a heterogeneous physics server environment, since we cannot have scripts detect which physics engine is in use, but as I understand the rumors the "mixed-physics grid" situation may not last a great deal longer? Thanks for looking into the issue! Yes, the double tap is a HACK work around. On the back end there are cases where we do the opposite -> push the avatar into the ground to force friction and prevent unwanted sliding. I thought this was the primary cause of this problem but ..... it appears "not quite", at best it is only one factor. To say this code is .... touchy .... might be an understatement. :)
I can understand that, sure... I've *yet* to work with any physics engine that was completely solid. I admit I've not had the budget to work with Havok, but I do understand that there are pathological problems encountered by all of them.
Thanks again for looking into the issue, it's very much appreciated! OMgosh! First I can kick people into orbit (way cool btw) but, not I can't even kick a person two inches away from me. This "issue" has hindered me greatly. Not only can I not kick, I can't hold king of the hill matches because of it. How am I supposed to have KotH matches without being able to get people off the hill? I've been waiting for how long for this to get fixed. I love Havok 4 but, fix the push already. GAH!
This issue seems to be stalled. There's been no visible progress, direct questions regarding progress are unanswered, and no sufficient information has been provided to allow for a workaround. The heart of the issue lies with the physics engine, so that is naturally the best and most proper place for a fix, but if someone could at least take the time to look into this just a little bit more from the LL side and give any ideas for a workaround (Kelly's previous suggestions failed miserably), I will implement it immediately.
I would like to see the same type of kick structure placed in Havoc-4 as we had before. The kicks used to work pretty good and the physics of them were fairly correct. Now its like kicking a fence post and you don't get the push back we used to get. Come on SL, lets get this back to normal.
Me too has noted what Malanchi, Aimee, Shindo (+ oth) reported. There's also a really minor side effect that may be suffered from this, which is being stuck in front of a kicker. By applying the proper physics, at least a determined distance is put among two opponents. A part from realism, it could be more equal for both, in case.
As an avid user of combat:si I can say that this is really not a small issue. It has a huge effect on the combat system because it kills some types of strategy and makes some very simple and predictable methods of attack more effective. The kick-push was a way to put a small distance between yourself and your opponent that prevented you from getting stuck and kicked to death. Now, after being stunned (this happens when you slash a block and during this time it is not possible to attack) the only option is to either block or run, both of which are sometimes riskier than taking a kick. I have a hard time believing combat:si is the only thing that's broken because of this. It's a huge issue and I hope it gets addressed soon. Note: this is not a huge push. This is a very small push that moves a few meters at most. The weapons are not designed for pwnage. They're designed for strategic dueling.
I don't see the issue here. Just increase the power of the kick. The system is changed to break abusive content. The SI swords should work fine with simply more power applied.
You guys should be complaining to the creators of them, not here. WarKirby, RobbyRacoon *is* one of the C:SI devs. He creates the Wave Katana and Taketori Katana and I'm one of his beta testers. So this issue *was* reported *by* a creator, not one of us lowly end users passing blame to The Lab. LOL
But yah, chiming in yet again cuz my original comment above is no longer valid. Kicks no longer push at all in Havok4 where my comment above said kicks were *either* ineffective or too powerful. Now kicks just do nothing. I realize the need for physics limiting griefer tools buz there *are* legit uses for push. Warkirby, increasing the power of the kick DOES NOT WORK CORRECTLY, as I said earlier in these comments. There are deeper and more fundamental problems with llPush under the new physics engine that simply cannot be resolved by "more power".
While simply adding more power does indeed have an effect, it is not a desirable one. The resulting avatar movement has nowhere near the same characteristics under Havok 4 as it did under Havok 1. It stutters, is nearly or entirely vertical, and does not suit the purpose for which the push was originally designed : To knock your opponent back approximately 2 meters, so they are no longer near enough to counterattack. Hardly a griefer tool, and it shouldn't be *completely broken* by a misguided attempt to intentionally break orbiters. I've just spent a bit of time looking through the list of push-related bugs in Havok 4, and more generally the class of bugs that under Havok 4 can best be described as "this product used to work, now it's broken". There are quite a few in that list, and this leads me to the compelling conclusion that Havok 4 is pathologically broken. And since Linden Lab has thus far categorically ignored several recent requests for status on these issues, I am left with little choice but to conclude that they simply don't care. We have thousands of active users of these products, and the best answer I can give to everyone who asks about these issues is that Linden Lab won't even give me a "we're working on it". . One more thing : If we are eventually forced - against all logic - to implement a workaround such as increasing the push power, it's easy to imagine that Linden Lab will eventually (even if unintentionally) discover the magic formula that will resolve the myriad push-related issues, and then the SI swords *will* have a griefer-level kick. That would then force us to update four different product lines, about twenty individual products, for more than sixteen thousand customers. That's not a trivial undertaking, and isn't something that any content creator would wish to do.
I believe I have a fix for the horizontal component of the push not happening. In short the friction keeping an avatar from sliding downhill was just too much for the push to overcome. We had this problem with forces from attachments before (among other things) and fixed it specifically for attachments. I have adjusted the model used for external forces to match how we do attachments (give or take some anti-gravity specific hacks) and it looks much better.
However, when comparing to havok1 the push is still only about 75% the force in havok4. Specifically in the script you gave me, if I adjust the version in havok1 to have a basepow of 15 (instead of 20) the pushes look about the same. Conversely increasing the basepow on the havok4 version to ~26 seems to be pretty close. Due to the nature of this function - how finicky it is, that I know there are some things with too much push and that it is already live in its current version - I'm tempted to just fix what I already have - the horizontal component change - and leave it with the reduced force. Thoughts? Curses? Flames? Fixing the horizontal component is really the highest priority for us, I believe, since that was the biggest problem I had with creating a consistent and reliable workaround; We can certainly deal with reduced force from our end.
THANK YOU SO MUCH FOR LOOKING INTO THIS!!!! That just made my day :) . Ok, I'm gonna move on then and consider this 'fixed internally'. I'll try and remember to update here when this code gets put on the beta grid to get more feedback, but watch the blog posts for it in case I forget.
This will *not* be in the next release (1.21, planned for this week hopefully), but should I hope be in the next server update after that. After another normal horribly bad day at work. I wish I had the word in my non-functional brain right now to express to you how happy this makes me. Just the information, and poss ability of the push issue being fixed puts a smile on my face a little more umpf in my step ^
kelly linden - 22/Apr/08 01:49 PM
"This will *not* be in the next release (1.21, planned for this week hopefully), but should I hope be in the next server update after that." So that means we should look for this possible fix in 1.22 server update, or the 1.21? This issue should be fixed in the current simulator version (1.21)
Arg, oops!! This is NOT fixed in 1.21, but will be in 1.22 (the next version we release, I hope).
Sorry Robby! This got caught in some pjira clean up I was doing. I thought I re-read all the issues to only get the ones fixed in 1.21, but not well enough I guess. :(
hehe... Well, I am still very glad to hear that it's coming :) | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Bug
Resolved
Major
Push avatar to avatar broken(Havok 4 Kick)
