[SDL] SDL 1.3 status ?

Mason Wheeler masonwheeler at yahoo.com
Wed Jul 27 22:24:49 PDT 2011

 See, I don't get that.  It just feels like it hasn't been thought through very far.  You make it sound like using OpenGL or Direct3D is something simple to transition to.  If anyone really thinks that, I invite them to take a look through SDL_render_gl.c sometime.  Look at how big some of the functions are, how much functionality is being wrapped in these simple SDL calls.  Most of them include at least 4 OpenGL calls; some have a lot more than that.  SDL takes a low-level API and builds powerful high-level abstractions out of it, and it does a good job of that.  Saying "but we'll only go this far, no further, and if you want advanced effects you need to master a completely new API and learn all about matrices and transformations and projections" is a bit off-putting.

If SDL doesn't provide rotation, to give just one example, then everyone is going to end up having to roll their own rotation abstractions.  And a bunch of them will be wrong, because that's what happens when people reinvent wheels.  And they're going to end up in games that say "made with SDL" on them, and people are going to look at the games and say "wow, SDL games suck! They've got graphical errors all over the place!"  When in fact, the problem isn't in the SDL part of the game at all; it's in the stuff that was left out of SDL.  But that won't matter; the perception will be there, and SDL will get a bad reputation for low quality.  Does anyone really want that?

From: Ryan C. Gordon <icculus at icculus.org>
To: SDL Development List <sdl at lists.libsdl.org>
Sent: Wednesday, July 27, 2011 9:55 PM
Subject: Re: [SDL] SDL 1.3 status ?

> Argh! Where are you people getting this from?

To be fair, you wrote a whole bunch about how the Linux desktop is irrelevant, with one sentence about software renderers tucked in before it. Smarter people than myself misread it, too.

More to the point: the software renderer isn't being held back. Sam and I spent a long time debating how much the renderer APIs should do...it wasn't really a question of how much does it make sense to implement in software, but how much of API should we be creating when people should really be using OpenGL or Direct3D. My argument was anything more than a simple blit is wasted effort, and once you add rotation, people will ask for every other piece of the fixed function pipeline.

So while the software renderer was probably a pain to build, it's built now, and it's not holding back further development, because adding more functionality to that API is a fool's errand.

We feel it's at the point where you can throw together a simple game with some basic effects quickly, and if you need more, you should use OpenGL.


SDL mailing list
SDL at lists.libsdl.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.libsdl.org/pipermail/sdl-libsdl.org/attachments/20110727/0c35ccb3/attachment-0008.htm>

More information about the SDL mailing list