It is true - the server could use improvements and they could both do with working better together. And this is something we want to do, eventually.
However these specific changes are about addressing some of the worst offenders - the blatently horrible cases where you fly off into the sunset because the sim or network hiccups or maybe even prevent you from spiraling into the ground when you drive across a border. Lets avoid designing or re-designing full-on client side physics and movement prediction here, for now. :) - Kelly On Tue, Oct 12, 2010 at 4:46 PM, Nexii Malthus <nex...@gmail.com> wrote: > Yeah, that is true. It should be able to detect the very source that causes > most movements. There might be issues on scripts that influence an avatars > movements via control events though, while that could be avoided by checking > if control events have been taken, it might misidentify scripts that take > controls but don't act on physics, such as AOs. > > I think the simulator could have a much better prediction system itself as > well, it shouldn't all fall on the client. > > For example if an avatar presses forward, the simulator should "jump" the > avatar forwards as if the avatar had pressed forwards back in time by their > average ping time. This would make it appear as if the sim is more > responsive. The client could also then simulate the forwards movement > instantenously. > > Modern FPS games have even smarter methods for synchronizing and predicting > physics accurately, such as tracking the recent positions of all players and > then 'correcting' recent collisions based on the players' connections in > realtime. > > - Nexii > > > On Wed, Oct 13, 2010 at 12:29 AM, Dahlia Trimble > <dahliatrim...@gmail.com>wrote: > >> Could checking to see if a user is still pressing a movement key fit into >> the algorithm? It seems it could be a useful input to help determine if the >> user really wants to continue along a dead-reckoning path. >> >> >> On Tue, Oct 12, 2010 at 2:51 PM, Dale Innis <daleinnisem...@gmail.com>wrote: >> >>> On Tue, Oct 12, 2010 at 5:38 PM, Simon Linden <si...@lindenlab.com> >>> wrote: >>> >>> > Some of my changes should help, however. The viewer will check the >>> edge of >>> > the region, and if there is no neighbor region, stops the object >>> motion. >>> >>> > Predicted object motion is also pinned to the >>> > allowable underground distance and top of the region. >>> >>> These will be major, and very welcome, changes to the SL experience! >>> >>> (Not to mention giving us oldbies one more class of "In my day..." >>> stories to tell...) >>> _______________________________________________ >>> Policies and (un)subscribe information available here: >>> http://wiki.secondlife.com/wiki/OpenSource-Dev >>> Please read the policies before posting to keep unmoderated posting >>> privileges >>> >> >> >> _______________________________________________ >> Policies and (un)subscribe information available here: >> http://wiki.secondlife.com/wiki/OpenSource-Dev >> Please read the policies before posting to keep unmoderated posting >> privileges >> > > > _______________________________________________ > Policies and (un)subscribe information available here: > http://wiki.secondlife.com/wiki/OpenSource-Dev > Please read the policies before posting to keep unmoderated posting > privileges >
_______________________________________________ Policies and (un)subscribe information available here: http://wiki.secondlife.com/wiki/OpenSource-Dev Please read the policies before posting to keep unmoderated posting privileges
