Re: [glob2-devel] Glob2 salvage proposal

[Stephane Magnenat]

Hi,

Andrew, I mostly agree with you. I also think that some part of the code 
should be rewritten. And because of that I take the opportunity to suggest a 
fundamental architectural simplification:
The network has been coded by Nuage and I thus don't know well its internals. 
The actual model is very complex because it is designed to pass through 
firewalls and NATs. As far as I know, we are the serverless game in the world 
that works best for games when all people are behind firewall/unconfigured 
NAT. But this feature has a huge price: a very complex and hard to maintain 
network code. Thus, the only solution I've found to make this part of code 
evolvable is to drasticly simplify it. And, the only way I've found to 
simplify it without loosing feature is to drop the UDP-based 
"game-runs-the-same-concurrently-on-each client" model in favour of a 
TCP-based client/server model. Indeed, the latter allows dedicated server, 
thus allowing people behind NAT to play easily. And the fastest way to do it 
(from a development point of view) is to use a thin client.
This was the motivation of my previous mail.

Now about the time; I will begin a new job at beginning of November and I thus 
don't know how much free time I will get. I should get more because it will 
be at 75% but I should get less because it is one hour away of my home...

This said, I really have to intention not to let glob2 die !

And before all, I'm really willing to discuss any suggestion, glob2 must be 
our game not my game.

Steph