So the motivation is to let those who use mac and linux play the game? I guess this makes sense, there should be quite a lot of them.
Out of curiosity, did you read the post I linked to?
Mac (and unofficial Linux) support are a fairly big deal (and the Mac point is probably the only one with the potential to single-handedly bring in some moderate amount of revenue), but there's a lot more on there than that
Re: .net 3.5 dependency. I think that .net 3.5 comes built-in to Windows 7 and there are more and more Windows users now that are using Widnows 7 If they are not prevalent among windows users yet they'll sure soon be.
Actually .NET deployment at the current time is a disaster, and a real embarrassment to an otherwise totally awesome platform. Yes, if they have Windows 7 they're probably ok. Probably. They won't lack the platform in that case, but a certain percentage of customers will have hosed .NET installs that will basically make it impossible to play AI War unless they want to go through the marathon of trying to uninstall/reinstall and/or truly repair a .NET install, or simply re-installing their OS (which can actually be faster).
Mono, I think is more like .net 2.0, not like .net 3.5
Yes, very much so; there's some neat stuff in 3.5 but 2.0 has everything we feel an actual need for. Very often we wind up having to write our own libraries for stuff to get the performance we need, so the difference is not great.
and .net 2.0 is installed with operating system almost on every windows os
This is simply not true
XP does not have it by default, and a significant portion of our customers have to install it when they want to first play the game. This problem will diminish over time, but it will take a while to reach "almost every".
and if not you can get it with windows update easily
Easily in the overall picture, yes, but it can take 2 hours and 2 reboots depending on the situation, which is a lot more than many folks are willing to go through to play a game.
One really killer point on the .NET thing is that even if the players have it, and the install isn't hosed, we still have no way of making sure (or really even knowing) that they have exactly the same versions and service packs as the people they are trying to play multiplayer with. Even a small discrepancy can lead to mysterious desyncs that force a reversion to a previous save (which sometimes will simply lead to another desync at the same or a different time). Granted, this is due to our networking model requiring 100.00% determinism across the parallel simulations, but that's the only way we can do it this side of Terabit ethernet
Do you think that if SlimDX ran on .net 2.0 you'd have a lot less problems with installation?
Possibly, but all the recent tech support issues we've had with .NET/SlimDX installs have been other things. And we've had an alarming number of those problems lately, which is one reason I'm really glad we're making the switch to a platform which inherently contains its prerequisites.
Lancefighter also mentioned another key motivation: having all our games running on the same engine. Helps no end in being able to use cool feature xyz easily in all the games that benefit from it.