GOD bless Jon and Client/server architecture vs ugly p2p

People dropping is just a result of a bad connection in some way or the other and has nothing to do with the difference between a synchronous and an asynchronous concept.
... or does it?!

SC2 shows off pretty good how p2p through a proxy can do pretty well without having to change as much as PA does.

not really, as its p2p. There is only one server and it hosts just the client. no game traffic or running the game.

if LAN will be possible without drawbacks.

This is completely and utterly wrong. Actually, the FAF forums alone have over 2500 (mostly active members). PA got backed 2 Million+ by over 44000 backers. I am pretty sure it will go huge when it releases.

And if you really want to see what a small community is capable of, I suggest you to take a look at competitive TF2. A community so small, valve basically just started to acknowledge its existance, after over 5 years. Yet, last year, 20000+$ were raised to send two American teams to compete against European ones at the i46 LAN and it's about to happen again with this year's i49. Not too long ago at a big completely community-created event, we raised 35000$ for charity on a single weekend.

You think the entire playerbase of PA won't be able to host their own servers?
Because I sure will if I have to.




...And how the hell could there be drawbacks to LAN.

I believe it has been confirmed that any random PC with PA installed will be able to host a multiplayer game. Just like FA.

I don't really see any drawbacks to LAN play. Technically you just give the PA client an IP to connect to and let the VPN take it from there.

Since there isn't much reason to encrypt game traffic you can just use IPSec encapsulation without encryption to reduce overhead or even use GRE at that point for even less overhead. Then again modern Intel CPUs all have AES hardware acceleration built in so its not like using encryption will incur any real overhead anyway.

Edit:

Actually now that I think about it I think DTLS would probably be the best choice for a game VPN because its based on UDP and won't add TCP's overheard to the game's own UDP based protocol like normal TLS would and it isn't its own IP protocol like IPSec or GRE which makes it much easier for people port forward it on their home routers.

Given the possibility of non-dedicated hosts this isnt a valid downside at all. Currently ALL players in an FA-game are basically doing the work of the server. If it were not FAF but instead PAF it would work just like it does now with the difference that clicking "create game" starts a server on your machine that the others connect to.

I think you too negative about what community might done.

There is other planetary-based game: Populous: The Beginning, it's never was popular like starctaft. But even now there is ~300 people who playing it and there is usually somebody in multiplayer on popre.net. Yeah, that game was released in November 1998.

If I play game for 5 years it's nothing to me to give 5-10$/month to support community server. Servers it's really cheap now.

So really, I never ever have any problem in finding server to play. It's usually problem to find people to play with. :lol:

I just want to make it really clear that no Unreal code has been anywhere near the PA source code. Epic tends to sue people who do that kind of thing so we have to be really careful about it.

Part of the confusion might be the UberNet back-end that we've built. We didn't have to redo this for PA since we already had it but it's a completely separate code base that was built to handle multiple games (there are currently 5 games running on the Ubernet back end right now).

I knew that much. What I meant to convey was that if you are playing in this asynch model and the server computer dies, then the game is done because none of the clients have the full picture. Similarly, this also exists in synch games because if one player disconnects the game isn't really the same anymore and might as well remake.
Sorry, I should have been clearer.

As programmer I understand that clearly, sorry if I make you worry about that.

I just point my own opinion about "how it works" because I worked in adjacent industry to game dev and have a bit deeper understanding that UE (and other game engines too) is thing-in-itself.

Some people think that game engines are independent from game code and custom modules, like it's easy to to port these parts from one engine to another. But in real world engines is like Qt or Boost, any part of them is completely unusual without other.

While I agree with the opening post I will miss crashing people with poor computers games by flying 100's of scouts over peoples bases until there pc melts and they discon like I did in supcom.

You are an evil evil person. I like you.

actually in supcom, them being specifically on his screen had no effect, it's the interactions with his units. Air units in supcom have a hight rate of fire and each projectile has to be individually simulated. If you had this battle far away from his base with someone else he wasn't allies with and also couldn't see; his computer would still melt down even though nothing was happening on his screen because supcom makes every single player calculate the ENTIRE battlefield with or without fog of war.

I enjoy the fact it's not based on apm that you can win.
I love to play Starcraft 2 but I'm in no way a match for really skilled players so I've not put down my barracks at 14 so they'll be able to steamroll over me.
In PA I can build my base as they can and then it's an issue of finding each other. It's very difficult to scout out an opponent early unless they are next to you. So you can really have that epic battle instead of losing to 3 units because you can't micro as well as they can.
It's been such a long day since those multiplayer rts of my youth when we'd build massive armies against our friends which is what I wanted to play again.
The best bit it's not yet got all the features in so it's only going to get better