SDL.NET vs FreeGLUT + DevIL + OpenAL

56 views
Skip to first unread message

Rob Loach

unread,
Jul 5, 2007, 11:04:28 AM7/5/07
to MonoXna
Since we're not going to be using SDL_mixer anymore, does it make more
sense to move from S DL.NET to another solution?

SDL.NET
- Image loading
- Window context
- Input
- Audio

VS

FreeGLUT
- Window Context
- Input

DevIL
- Image Loading

OpenAL
- Audio

............. In a related note, I talked with the Jonathon of the
OpenAL.NET project and he said that he'd give us as much support as he
can. Great news!

Rob Loach

unread,
Jul 5, 2007, 11:48:41 AM7/5/07
to MonoXna
Just a quick note that I'd personally like to stay with SDL.NET if we
can.

Stuart Carnie

unread,
Jul 7, 2007, 6:50:33 PM7/7/07
to MonoXna
I too would like to advocate SDL / SDL.NET. SDL is under development
with a number of enhancements, which SDL.NET can expose. Can't
remember specifics, but it was certainly of interest to this project.

Cheers,

Stu

Antti J. Oja

unread,
Jul 8, 2007, 1:20:23 PM7/8/07
to MonoXna
Greetings!

I don't personally agree with the usage of either SDL or SDL.Net on
this project. They do make some tasks easier, but I don't perceive SDL
as a very fitting option for a project of this magnitude. It will
probably end up being very constricting in terms of options or
alternative implementation approaches we can take later on. I neither
like Glut, as it is - as Alan once pointed out - rather antiquated per
se, and favored for small demos rather than full-blown frameworks.

Right now, I'm implementing a Mono-port of a commercial rendering
engine. I chose to use Glfw (and OpenGL with that, obviously), OpenAL,
DevIL, Cg and some other minor libraries for this job. Glfw is
magnificently simple to set up, doesn't require X11 on OS X, and is
much lighter to use with DevIL (and a bit of managed code) than
any .Net-bridged SDL I've seen thus far. Glfw just provides simplified
window setup, attachment of rendering targets etcetera. In other
words, it doesn't get in the programmer's way while he's doing the
actual implementation.

It's a very short argument against SDL(.NET), but hopefully it
suffices to say that given the example above, SDL doesn't scale well
for performance-critical solutions. If anyone dares to say Mono.Xna
isn't performance-critical, that is untrue. Microsoft has "cheated"
with their Xna, mostly because it's a hybrid of managed and unmanaged
code. They have the advantage, and YET XNA isn't very fast.

Cheers,
A.J.

Stuart Carnie

unread,
Jul 9, 2007, 1:43:02 PM7/9/07
to MonoXna
I'm cetainly still open too - it will require some refactoring for me,
to update the graphics classes, but no big deal.

Are you implying we use GLFW through the Tao.Glfw bindings?

Let's discuss the media frameworks on the call.

Cheers,

Stuart

Stuart Carnie

unread,
Jul 12, 2007, 8:45:44 PM7/12/07
to MonoXna
I did some experimentation and review of Glfw, and am definitely
leaning towards this, since it has some functions to closer map to
Xna, that was missing from SDL. In addition, OpenAL may be a better
fit for the Audio layer.

Cheers,

Stu

Alan McGovern

unread,
Jul 13, 2007, 3:31:41 PM7/13/07
to mon...@googlegroups.com
Well, i know very little about this sort of thing, so i'll go with whatever is decided upon. I don't really have any expertise to add to the discussion ;)

Alan.

> > > I too would like to advocate SDL / SDL.NET .  SDL is under development

Stuart Carnie

unread,
Jul 13, 2007, 7:48:50 PM7/13/07
to MonoXna
I also like the fact that the Glfw window doesn't change the mouse
cursor, it leaves it as the standard OS cursor... ...which is the same
as Xna.

I'm eager to try the joystick API tonight when I get home, as it may
also be dynamic, in that if you attach a joystick, it will recognise
the change. SDL does not. It is only initialized upon start up.
This behaviour may also be closer to Xna.

The keyboard support may be better too - functions such as disabling
system keys (useful for full-screen mode).

I'm going to play a little more over the weekend, and locally
experiment with a Glfw version.

Cheers,

Stu

On Jul 13, 12:31 pm, "Alan McGovern" <alan.mcgov...@gmail.com> wrote:
> Well, i know very little about this sort of thing, so i'll go with whatever
> is decided upon. I don't really have any expertise to add to the discussion
> ;)
>
> Alan.
>

> > > > > I too would like to advocate SDL / SDL.NET. SDL is under

Stuart Carnie

unread,
Jul 13, 2007, 8:11:41 PM7/13/07
to MonoXna
Kill that idea...glfw doesn't support joysticks on OSX - even in 2.6
beta 1. I looked at the source and it is entirely stubbed out.

I will do some more research over the weekend; however, it appears
that SDL is still the way to go right now. Nothing still precluding
us from using OpenAL, but we'll see..

Cheers,

Stu

Antti J. Oja

unread,
Jul 27, 2007, 6:25:01 AM7/27/07
to MonoXna
Actually, having spoken to Camilla Berglund who maintains GLFW, she
claims that the foreseeable 2.6 release will contain gamepad/joystick
support for OS X.

Stuart Carnie

unread,
Jul 27, 2007, 2:27:25 PM7/27/07
to MonoXna
I think we should still continue with SDL as my opinion is that it is
more mature across all the platforms we plan to support.

I honestly don't think it will be too hard to change between the two
later, given Tao.Gl and Tao.OpenAL will probably be the largest touch
points between our Xna implementation and the hardware.

Jérémie LAVAL

unread,
Jul 30, 2007, 11:50:00 AM7/30/07
to mon...@googlegroups.com
Hum, SDL is still in a lot of place from Graphics (windowing, image loading and such), the whole Input namespace and some other miscellaneous stuff (maybe Content Pipeline). IMHO changing toolkit later will be a pain wherever we choose GLFW or SDL. The unique solution to this would be to use abstraction over concrete implementation built with different toolkits (maybe a IoC pattern would fit there?). The cons of that way is that it adds an extra layer to Mono.Xna (when you know that SDL.NET already weights a lot).

My personal taste goes to Glfw because of the points Antti already expressed (simpler, lighter...) but then it would mean a rewrite of what he is already in svn. Personally it's not bothering me that much (we don't have a stressing corporate agenda) but then I know that Rob (for example ;-) ) wouldn't like that idea.

Stuart Carnie a écrit :

Stuart Carnie

unread,
Jul 30, 2007, 12:04:22 PM7/30/07
to MonoXna
I will touch base with the project team over at Glfw and find out if /
when we may see the joystick implementation and how complete the
library is for the primary platforms (Linux and OS X, given Win32
technically has a complete implementation).

Cheers,

Stu

Reply all
Reply to author
Forward
0 new messages