Real Time Network
by Jarred Michels · in Torque Game Builder · 09/06/2006 (4:09 pm) · 10 replies
We really need a real time network for the TGB this will bring a hell of alot more people to GG. I think every game should have real time network to me that makes a game alot better.
About the author
#2
I for one thing real-time networking would be a great thing to have, but having little networking experience myself, I don't know how feasible that really would be.
09/07/2006 (8:33 pm)
I'm rather confused by your post, David. It sounds like you're being sarcastic, and making Jarred's suggestion out to be ridiculous, but I can't quite tell.I for one thing real-time networking would be a great thing to have, but having little networking experience myself, I don't know how feasible that really would be.
#3
A cursory search of the forums shows that this topic comes up weekly, and the simplification of the network model to elevate ease-of-use has been well stated. I believe that the terms real-time and turn-based are a referential problem which often comes down to confusion with people who think in terms of games that they have played that are turn-based or "real-time" versus network data transfer and figuring out how and when to transmit your data. for a fast-action multiplayer game with pin-point accuracy, the networking model may not be sufficient. With some "real-time" strategy games which process major section and determine when significant pieces need to be adjusted, the network model will work as long as you plan and playtest the update model. Is it as robust as TGE's network model? No. Will it cripple your multiplayer gameplay? Perhaps. It depends on your multiplayer gameplay.
The third paragraph was my thoughts on a content pack for advanced networking, but the split between engine models (TGB and TGB Pro) makes a pack more difficult to plan for and release. So I touted that as a bad idea, or at least an idea for another day.
Currently, using TomB's resource to combine TGE and TGB Pro is the best way to go if you need more robust networking than what is provided. And he is also extremely knowledgable about what TGB and TGE can and can't do in terms of networking.
The poor guy's also entirely too knowledgable about the internal details of Constructor's Undo system so he doesn't get much sleep.
09/07/2006 (8:58 pm)
I'm a rather schitzophrentic poster. The first two sentences are tongue-in-cheek to what would make my artwork better (not having me do it).A cursory search of the forums shows that this topic comes up weekly, and the simplification of the network model to elevate ease-of-use has been well stated. I believe that the terms real-time and turn-based are a referential problem which often comes down to confusion with people who think in terms of games that they have played that are turn-based or "real-time" versus network data transfer and figuring out how and when to transmit your data. for a fast-action multiplayer game with pin-point accuracy, the networking model may not be sufficient. With some "real-time" strategy games which process major section and determine when significant pieces need to be adjusted, the network model will work as long as you plan and playtest the update model. Is it as robust as TGE's network model? No. Will it cripple your multiplayer gameplay? Perhaps. It depends on your multiplayer gameplay.
The third paragraph was my thoughts on a content pack for advanced networking, but the split between engine models (TGB and TGB Pro) makes a pack more difficult to plan for and release. So I touted that as a bad idea, or at least an idea for another day.
Currently, using TomB's resource to combine TGE and TGB Pro is the best way to go if you need more robust networking than what is provided. And he is also extremely knowledgable about what TGB and TGE can and can't do in terms of networking.
The poor guy's also entirely too knowledgable about the internal details of Constructor's Undo system so he doesn't get much sleep.
#4
All I want is for "basic" real-time networking. For example, if I move a sprite on the server it will also move on the client. Also if a collision occurs on the server, it will occur on the same *exact* spot on each client.
*sigh*
I'd even be willing to pay $50 or so for a solid real-time networking add-on.
09/13/2006 (1:09 pm)
I've been begging since I bought the first early adopters release (beta 1, if I remember correctly) for real-time networking.All I want is for "basic" real-time networking. For example, if I move a sprite on the server it will also move on the client. Also if a collision occurs on the server, it will occur on the same *exact* spot on each client.
*sigh*
I'd even be willing to pay $50 or so for a solid real-time networking add-on.
#5
09/13/2006 (1:15 pm)
@John: the problem is however that there isn't any way for it to be "basic". Your second request (if a collision occurs on the server, it will occur on the same *exact* spot on each client) in fact isn't even possible across multiple, different platforms working through the Internet.
#6
I'm struggling with this in my Scorched Earth game clone. I thought that TGB TBS (turn-based) networking would be enough, but I'm having difficulty in having the projectiles follow the same trajectory (and hit the same spot on the terrain) on everybody's client. :(
09/13/2006 (1:24 pm)
Sorry, maybe I asked for the wrong thing. I'm still a bit unsure about why every PC doesn't perform calculations the same way (eg. rounding errors, etc), but in a multiplayer game environment everything needs to be sychronized. So for example if a ball hits the ground and rolls to the side before it stops, the ending position should be the same on everybody's game. Otherwise some people would have the ball a little too far to the left, or too far to the right, etc...I'm struggling with this in my Scorched Earth game clone. I thought that TGB TBS (turn-based) networking would be enough, but I'm having difficulty in having the projectiles follow the same trajectory (and hit the same spot on the terrain) on everybody's client. :(
#7
09/13/2006 (1:33 pm)
Have you looked into using pre-generated tables to define your trajectories, beginning, and end-points on all clients? Pre-calculated tables were a huge boon to developers back in the day and could help you in this instance. Catching things mid-trajectory (such as a ship that flies in the path of our poor projectile) might be a problem, though.
#8
I'd actually like to add "black-holes" and stuff like that to the game (and anti-air guns, etc).
The built-in T2D physics would work great if it were more stable (but I guess it's good enough for now) and networkable.
I'm open to any other ideas though!
09/13/2006 (1:56 pm)
Yeah, the whole mid-trajectory thing is the problem.I'd actually like to add "black-holes" and stuff like that to the game (and anti-air guns, etc).
The built-in T2D physics would work great if it were more stable (but I guess it's good enough for now) and networkable.
I'm open to any other ideas though!
#9
A real time networking is really a minimum ! After all you done with TGB is that so impossible ? What i don't understand is why so much of people need real time networking and why with such a good community NOBODY as release a real time addon ! (and no tweaking with TGE is NOT an addon to TGB).
Well, i will try with "commandToClient/Server" but i'm pretty sure i will get Huge Latency.
Another solution would be offering the abilities to write is own UDP TCP/IP socket with TorqueScript !
When reading the TGE award winning networking features i really don't understand why in 2006 we're still stick with turn based game...
10/01/2006 (5:45 pm)
Well i don't remember the last AA game i've play in turn based mode...A real time networking is really a minimum ! After all you done with TGB is that so impossible ? What i don't understand is why so much of people need real time networking and why with such a good community NOBODY as release a real time addon ! (and no tweaking with TGE is NOT an addon to TGB).
Well, i will try with "commandToClient/Server" but i'm pretty sure i will get Huge Latency.
Another solution would be offering the abilities to write is own UDP TCP/IP socket with TorqueScript !
When reading the TGE award winning networking features i really don't understand why in 2006 we're still stick with turn based game...
#10
10/01/2006 (6:35 pm)
@Rivage, TCPObject() is a TCP Socket object ... and as far as networking 'addons' go, the TGE networking to TGB resource is an 'addon' and is more then suffecient if you need more networking. As for GG, they've got there hands full doing quite a bit -- Torque Network Library is opensource, www.opentnl.org/. 
Associate David Montgomery-Blake
David MontgomeryBlake
I would have to disagree about every game having networking, let alone "real-time" networking. Though I also really think that the term "turn-based" networking isn't right for TGB. Of course, I also can't think of another user-friendly term. Not every, and actually not many, need up-to-the-moment network ghosting. Some definitely do, but even many RTS games use client simulation and relational information in burst cycles to accommodate for people with extremely slow network connections.
I would think that an Advanced Networking pack would be a good idea. The core reason being that by pointing out that it is advancd, you will have to have much more experience with networking than the simple "it works" networking currently in TGB. The more robust the model, the more experience needed to effectively use it. Of course, I have no idea how it would work with the no/source distinction between TGB and Pro.
Okay, bad idea. I swear my next one will be better.