What are the pros and cons of different libraries and programming languages for 3D game developement? My idea from what I have found is this:

---

XNA:

pros:
-easy to use
-a lot of thing already done

cons:
-only for windows
-possibly little less powerful
-bloated?

---

DirectX:

pros:
-Compared to OpenGL, a lot of things (loading models, etc.) is already in the library.

cons:
-Only Windows

---

Allegro/SDL+OpenGL:

pros:
-cross platform
-control over the resulting code

cons:
-requires a lot of programming.

---

Which library would you choose for a commercial 3D game? Why? What are other possibilities?

XNA.

A major pro for OpenGL is the stable API. Beyond that I think you have the pros and cons correctly listed. There are definitely ethical issues with XNA, but do you care?

EDIT: XNA titles cannot be used on Mac computers unless the owner has also installed Windows. Macs account for 20% of US consumer sales nowadays (source).

Mac is only a marginally larger gaming market than Linux.

Subjective viewpoint is subjective.

Windows may be the largest target for commercial development, but that's a far cry from being the only one that matters. Using open APIs like OpenGL will make it so it's easier to port, if you want to support other systems now or in the future. And even if you don't want to support other systems, using those same APIs will make it easier for people on other systems to run it in something like Wine (ie. more sales).

Do realize that many many people would gladly ditch Windows if the games were (competantly) ported to their OS of choice. They only stick with Windows (usually dual-booting) because people keep developing exclusively for it. It just perpetuates itself.. people use Windows because apps/games aren't being ported, and apps/games aren't being ported because people use Windows. Even if you don't want to port, using more open technologies will help break that cycle.

XNA should work in WINE, it's just a question of time.

Yes, but for example, not using XNA requires you to do more work to get it on xbox if you choose to do so. Most "gamers" out there aren't running Linux, therefore you have to weigh the benefit of selling a few more copies to Linux folk vs. using a framework that allows instant use on Windows and xbox.

Also, XNA is all .NET, and given Mono's success thus far, XNA shouldn't be all that hard to add layered over opengl. So in that regard, its a good choice as well.

Yes, but when you weigh the benefits you get from it, and the potential speed difference from you reinventing the wheel to layer similar stuff over an unmanaged language, it is beneficial. I've been writing in C++ (most C++ I've done in a while) doing this PSG project. And I do have to admit, I miss a lot of the extra syntax goodies and shortcuts you get in C#.

No it shouldn't. XNA is managed, so it would work under Mono, not Wine.

They do have a share, but is that share large enough to potentially spend extra resources on? If it costs you enough extra money to develop using open APIs that you won't recoup that cost plus an extra profit to make it worth the time, what is the point in doing it?

I'd say there are relatively few gamers on platforms like Linux and OS X, much less than that of Windows. Mac users are stereotypically naive and not afraid to spend money, so you can make some good money porting your games and selling them for a nice profit (look at the Mac ports of a few year old games that still cost $40 or $50, whereas the PC version costs $9 or so). But, how many Linux users do you see going out to buy software of any kind? Most of the time, the Linux user will pull what they want from a software repository. I'm just saying, Linux isn't a platform that is likely to be worth the hassle by the time you deal with the annoyances of distributing binaries (opensourcing your commercial game is not as feasible an option), changing APIs or writing layers so you can still keep your stuff on closed platforms with different API, and such.

Linux isn't ready for mainstream gaming, still having crappy drivers, and hard to gain market share. I don't see very many stereotypical gamers changing over to Linux just cause games are released for it. And most stereotypical Linux users aren't into games.

Support for Mono in XNA has not advanced since July 2007. That's 15 months ago.

How closely have you looked at the boxes? EA will reputedly ship as many games as possible on dual format PC/Mac discs from now on. Though so far, I'm only aware of one (ignoring all the Sims expansion packs).

To be fair, I think most users of Linux and Windows aren't into games. I think most users of Linux are interested in web serving 24 hours a day, and most users of Windows are interested in using Outlook and Office between about 9 in the morning and about 5.30 in the evening.

I don't think anyone will dispute that, irrespective of whether that makes it relevant to someone producing games, Mac OS is the OS for with the highest proportion of users are people that personally own the machine and use it for home-based desktop activities.

If you were planning to sell your game, it may be relevant that while the Mac accounts for just 20% of US consumer sales by volume, it makes up 35% of dollar value.

Most gamers don't consider open source games to be "real games".. they're more novelty time-wasters, if that. Sure, there are some gems (eg. Nexuiz), and there are some good commercial games (eg. UT2k4, Doom3, NWN), but it's nowhere near the level it is on Windows.

PS. And when I say "games", I mean shiney new AAA titles with omgprettygraphicsandphysics. Something gamers get hyped about and look forward to.

EDIT:
For example, can Linux users look forward to FarCry 2? What about Legendary? Or how about Diablo 3? Starcraft 2? Fallout 3?

I've spent very little time looking into XNA but there is a lot about it that seems easy to use and attractive, epecially for the indie games market. It certainly lowers the cost of entry for first timers and indies, but it does lock you in somewhat.

With that caveat, what stops me from using it:

• I'm primarily a C++ programmer and despite the benefits that C# offers am not yet motivated to learn it beyond trivial uses so far. A good engine/library I know gives me all the head start I need. I'm not convinced after the initial boost in productivity that when one gets into the nitty gritty of a complex title that XNA is going to provide that much benefit.

  
  

• Limited to Microsoft Platform. I'm writing commercial games that can be targetted to the console, and have not yet decided I want to be microsoft specific. XNA limits me to Windows and Xbox 360, I'd quite like to be open to PSP, PS3, DS, Wii, Pandora or whatever at this point, and the business side of things will effect my decision. (All those other systems tend to be very linux like in the way they do things, if not actually based on linux, many linux based tools etc)

  
  

• A game written with XNA will most likely lead you down the XNA publishing path which has lower royalities than a self published title going through the normal publishing route. I'm not 100% up to date with this though but it certainly used to be true. There is going to be a lot of competition in the XNA arena too, as I've already stated there is a lower barrier to entry.

For the hobbyist game writer, maybe. But commercial game developers don't care if it is free or open, they are willing to pay. And by well supported, they don't consider web forums good support.