CS:S is terrible

[Cheeseh]

hmm never tried hl 2 multiplayer .. ? Can't remember how to use it
you gotta make some compromises though, man.. how would you get extremely nice physics across a network (real time problems and there are limits on the amount of data that can get sent)

[MarD]

Heyyo,

Heh, I dunno, HL2 coop worked fine.. until it crashed. I guess they never completed the HL2 networking code since they dropped the multiplayer side of the project (well, for the time being at least...).

What happened is I died from 1 explosive barrel, followed me bud, he blew me up with another by whipping it right at me... and then I died, and it crashed seconds later.

[Pierre-Marie Baty]

"how would you get extremely nice physics across a network"

Prediction. Client-side prediction.
The server only tells the clients about the impact parameters. Entities involved, origin, velocity, date, and server's random seed state. If the physics are 100% well-made (i.e, completely FPS-independent), all the clients should end up in the exact same state, no matter their machine or net connection.

[dead bwoy]

hmmm, I have noticed that clients screens end up different in CS:Source.
What I mean is, say I get naded... On my screen (client), I get blown up on top of a crate. And on my room-mate's screen (client), who is watching me, my body ends up on the ground next to the box.

So I guess what you're saying, PMB, is that the client-side prediction needs to be improved to be able to use HL2 physics in CS:Source?

[MarD]

Heyyo,

Heh, yeah. I think they dummed down the physics in CS to lower the sizes of packets. For the dead bodies thing? that's a great example of how they diddn't fully implement the physics calculations into the game, to lower packet sizes. Same is done in HaloPC. When bodies die, they don't sync them so that's 1 less thing the game has to worry about.

[dead bwoy]

Originally Posted by MarD

... they diddn't fully implement the physics calculations into the game, to lower packet sizes.

Was this done to still allow 56K users to be able to play on-line?

[Exilibur]

well, since only players and bullets can interact with objects affected by physics, why would they need to send any more information?

I understand that each machine of course has to make the exact same predictions about the behaviours of each item affected by either players or bullets, but i don't really understand why this should be such a problem.

My logic tells me that if all the clients have the exact same information, then all the clients would behave identical. This isn't quantum physics...

I understand that I must have missed something, since this isn't already working as I predict it should be able to... but i can't see what. Can someone enlighten me?

[Huntkillaz]

when cs:s was in beta stages, i had a good play test of it, and i found that it was possible to jump on objects like barrels and stuff but when i ran into them i'd bounce or go right through the object this was around aug 29

so it seems that the latest updates may have changed this?

[sPlOrYgOn]

I played the beta when it barely started and it's exactly like how pmb said it was except the beta was only on de_dust so no hosties..

[dead bwoy]

You can stand on the props, but not very well. You jiggle around everywhere, it'd be like shooting from a ladder!
There are some console commands I've yet to play around with...
something about pushaway_force and a couple others...
I'll see if I can try this out sometime tomorrow.