Ask Slashdot: Should Valve Start Their Own Steam Linux Distro? 316
Duggeek writes "There's been a lot of discussion lately about Valve, Steam and the uncertain future of the Windows platform for gaming. While the effect of these events is unmistakably huge, it raises an interesting question: Would Valve consider putting out its own Linux distro? One advantage of such a dedicated distro would be tighter control over kernel drivers, storage, init processes and managing display(s), but would it be worth all the upstream bickering? Would it be better to start anew, or ride on a mature foundation like Fedora or Debian? Might that be a better option than addressing the myriad differences of today's increasingly fracturing distro-scape?"
Comment removed (Score:3, Interesting)
Re:Just what the world needs (Score:5, Interesting)
Indeed.
When I started with Linux, it seemed the choices were few: Slackware, or Yggdrasil (Red Hat, Suse, and Debian were a few years hence). Matt Welsh's fabulous book "Running Linux" focused on Slackware, and so did the rest of the Linux Documentation Project (is the LDP even still alive?). a.out was still a viable, and used, executable binary format.
Package management was shit: You installed a new package on your existing system with (at best) a "./configure&&make&&make install" as root (WTF is sudo?), ran ldconfig, fixed whatever it broke, and moved on.
Today, there are a myriad of safe (and unsafe) choices. And while the capitalist in me says that choice is good, the pragmatist in me says that it's really a burden.
The reasons for the crop of shit that we've grown are obvious: There is an incongruity between the folks who want to pay for an OS (Red Hat), the folks who want a free (libre) OS (Debian), folks who want an efficient OS (Gentoo FTW), and folks who want an OS that Just Works (Ubuntu).
So I'll be the first to say it: Yes, the community can stand to have a distribution wherein games Just Work. Because in having games Just Work, it's likely that proper low-latency audio will also Just Work. And from there, it's easy to have video Just Work. And at that point, it starts to sound a whole lot like what BeOS was...except it's still *nix, and it works on modern hardware.
Does it route packets? Does it run VMs with seamless precision? Can I do backups on an ancient Travan drive using ftape? Does it speak Arcnet or Token Ring? Who cares! Seriously. (I write this as a geek who has done all of these things, with a love for computing history, who has a thermal teletype, a box of paper, and a dedicated spot in the living room with suitable wire already installed, just waiting for a modernly-useful application that would benefit from such placement, as opposed to the dual-core 1.2GHz Linux box that I carry in my pocket.)
What the world could use right now, in my humble opinion, is a free(ish) OS that can do useful things with games media with great expediency and reliability.
Why?
Traditional user applications have run so fast ("faster than instantaneous" as a someone once told me is a bit of an exaggeration, but does fit with the current user experience) on any new hardware for nearly a decade that it's silly to even consider them as a goal. For all we complain, both Firefox and Open Office work fine even on rather ancient hardware (for instance).
Scientific applications increasingly rely on GPU calculations which rely on drivers for video cards which are primarily written for gamers. And as a scientist, one shouldn't need to care of the OS is totally free (libre), but whether or not the math is good and fast.
And server apps, well...gosh, Linux has done that very well since nearly day 1. The market needs no relative improvement in this area. It's nailed.
So a focus on low latency, for both video and audio, is a boon for gamers. A focus on making modern graphics, sound, and input hardware work well (through driver and API improvements) is a boon for both gamers and the scientific community. Give these goals a profitable shot in the butt by making games snappier than on other systems, and the rest of the demanding applications that common consumers actually use (AV production, graphic arts, fucking Youtube/Facebook/et al.) will happen naturally -- while also benefiting the rest of the users in the scientific community, and maybe (but not likely) in the sever realm.
(The above is just a dream from me, a random dude, who has used x86 computers for a couple of decades.)
Re:why on earth would they want to do that? (Score:5, Interesting)
Linux is a FOSS operating system. There is more to the question of should Valve start their own distribution. There is how much input they want into that operating system. From examples like what Google is doing with Android with a layer on top to more like Ubuntu focusing on services and support as well as ease of installation, to simply branding. Taking an existing distribution and contributing funds for it's development and just changing branding.
Then you can look a working through the various levels, starting at branding and getting market exposure and working up to a fully internally developed version with a gaming layer on top and ensuring that gaming layer is compatible with the majority of games you distribute. You can even look at making your layer able to work in parallel with Google's Android layer. The real advantage of FOSS you're not forced down one companies lane for the benefit of that company, you can choose a full range options and retain control of those choices.
Valve of course is not really likely to produce a gaming console and far more likely to produce a specification for a gaming console and allow manufacturers some scope of individuality in development and manufacturing of the console. The principle being to take M$'s profit (the windows and xbox tax burden) and distribute it amongst a far wider market.
Re:Just what the world needs (Score:4, Interesting)
They don't need a distro... (Score:3, Interesting)
Perhaps what Valve need to do isn't create a replacement distribution of Linux, but simply a replacement interface for it. Ditch X11 and all its window management software, and just run it all inside a Valve-designed user interface created to make things nice and simple. They could create a UI with consistent and familiar rules, publish API's to allow developers to create applications that use Valve's hardware-accelerated and streamlined system natively, and allow X11 to be run alongside this new primary user interface just like any other application.
On second thought, I could swear I've heard of something like this before...
Re:Just what the world needs (Score:4, Interesting)
It's a matter of using the right tool for the job. Valve will need to set some standards for their Linux environment, so that developers know what to expect on their customers' machines.
Creating their own distro or targeting an existing distro will accomplish this, though in a rather ham-fisted way (they seem to be going with option 2 atm, targeting Ubuntu). A better alternative may be to define a set of libraries and let the distros create meta-packages.
Re:Just what the world needs (Score:3, Interesting)