jmke

"SteamOS has been further inspected to see what kind of technical solutions it uses. The Debian-based OS uses Linux 3.10, shipping with a heap of patches applied, with the most focus being on real-time-like features. The kernel is also using aufs and they seem to be sitting on some bug fixes for upstream on top of that. The kernel is not using the new Intel P-State driver, with the reported reason being, 'it causes issues with sound being choppy during BigPicture trailer video playback.' SteamOS is using SysVinit as its init system. The desktop is backed by X.Org server 1.12.4 and a custom desktop compositor which seems to be a 4,200-line patch on xcompmgr. Catalyst and Mesa components can be found on the system, but so far only NVIDIA is officially supported. The system boots into Big Picture Mode, but the user can drop into a GNOME desktop. Responsible for a great deal of the kernel changes, SteamOS compositor work, and other SteamOS code is Pierre-Loup A. Griffais, a.k.a. 'Plagman'. He was a NVIDIA employee dealing with their Linux support. Another Valve employee doing lots of the SteamOS system-level work is John Vert, who up until last year was a Microsoft employee since 1991. There's also other former Microsoft employees on Valve's Linux team, like Mike Sartain.

http://linux.slashdot.org/story/13/1...ood-of-steamos

__________________

jmke

as usual, slashdot excels in their comments section, with this great piece of writing regarding what SteamOS is, and why it exists and will be a big deal in the years ahead:

SteamOS is an OS designed not just for gaming, but for a specific subset of gaming - using a controller and television instead of a mouse, keyboard, and monitor. The UI needs to be significantly different. You know how everyone bitches whenever an OS tries to reinvent the UI so that it works on both tablets and computers? This could have been the same situation, but Valve was smart enough to realize "hey, nobody wants to use a 10-foot UI on a 23" monitor, and nobody wants to type with a controller when they have a keyboard. Instead of pissing off our existing users *and* alienating the new target audience by making a compromise that fails at both, let's have two completely separate modes".

That's what SteamOS is designed for - a difference user interaction method. Or to be more precise, that's what Big Picture Mode (the Steam mode that SteamOS boots to) is. Big Picture Mode can be enabled as the default on any Steam install (Windows, OS X or Linux), and it's relatively simple to get Steam to launch by default as well.

However, SteamOS includes more than just a few default UI settings. There's the incredibly simple installation script - it offers very little customization, but it requires almost zero knowledge outside "getting your computer to boot off media instead of primary disk". That's essential for this particular niche, but would you want Debian dumbed down like that?

Or the stripping of unneeded crap. As I read TFA, I learned they built a rather customized compositor focused on game performance. Doesn't work too well in windowed mode, but it works well for fullscreen with UI overlays. Does that sound like something Debian ought to use?

Same for their kernel tweaks (some realtime scheduling stuff and disabling things that caused bugs with games), or their stripped-down install, or the dozens of other changes people are still trying to find.

But here's the thing - they are making almost all of this available as patches. It's open-source, except for Steam itself and the improved proprietary drivers. If Debian sees a use for these changes, they can merge it in. But to counter your inevitable repetition of "just make it a patch ****-gargler", you need to look at Valve's logic.

They saw Windows 8, and they were afraid. They realized that as long as PC gaming was reliant on one company (Microsoft) for an essential component, and that company has not just apathy towards PC gaming, but an outright reason to try to kill it in favor of their higher-profit-margin console, no matter how well Valve did at making games or keeping Steam running, their business could be destroyed. And with the Metro stuff and the locked-down app store, they saw a direct death threat. Shortly afterward, they started pushing their Mac port harder, and started work on the Linux port.

Given that history, would you really expect them to make themselves reliant on someone else for their console? Their thinking is basically "if we control our own OS, even if every other OS maker turns hostile to our market segment we can still keep it running". They have no problem with people running Steam on Windows or Ubuntu or Debian or ****ing Slackware for all they care - but they want to make sure that there's at least one OS that will *always* be there to play games on. Hell, they even recommend Ubuntu+Steam for the old desktop experience, not SteamOS. Nobody is going to be rebooting into SteamOS - they'll be running it as the only OS on the machine because it's the only one that's usable on that machine's setup.

But everything I just said was kind of pointless, because you don't understand the very issue you're bitching about. Fragmentation on Android is a problem because programs that use new features do not work on old versions. Fragmentation of window systems or other APIs is bad because you have to write a new version for each system. And desktop Linux dislikes fragmentation because it makes it harder to evangelize the platform when there's infighting.

SteamOS specifically avoids all that. As far as Steam is concerned, "Linux" games are "Linux" games, and run on Ubuntu, SteamOS or any other platform that runs the "Steam runtime", a basic compatibility layer so games can assume the existence of certain things.

They aren't fragmenting compatibility. If anything, they're reducing it by making a standard for games to work against.

__________________