I'm reading these forums to make up my mind on buying DGSDK or maybe not, and I find this...
Ladies and gentlemen, a little piece of advice: don't upgrade unless its ·absolutely· neccesary.
I won't be talking about those poor guys who were developing on win2000 when XNA kicked'em off. But think about this: Last year I was using an old (ancient?) Cray X-MP2 which gave us ONE error after TWELVE years. It was "upgraded" (grin) to a SGI-signed big box... After two weeks of "fighting-against-da-bugs" (it was like Starship Troopers, doh!), we got the boss to bring us our old Cray back.
Same goes to installing every DXSDK update, not to mention the problems you'll encounter on "fresh installs" of these DXSDKs (missing DLLs, to mention one).
As long as I "choose" to use DX, I'll try to use as much ANSI C/++ as I can, AND I won't be "upgrading" to VC++2005 just to find I must rewrite half of my code because of M$ "beta-versions-labelled-as-ready".
Which takes me to the point I doubt whether to buy DGSDK or not...
How much of it DOES really work? Will we get pointers to objects anytime soon? How can users give a hand to its TWO programmers? Does transparency work (objects and PNG's alphe channel)?
Perhaps I should have posted some of this to a new thread...
BTW, don't tell my boss I told you about the Cray
"Never give up! Never surreeendeeer!"