A Message From Valve's Gabe Newell

[geamandura]

Regarding Icculus: I don't believe that he actively perpetuates trying to convince companies to bring their stuff to Linux. Instead, he just seizes the opportunity when a company does accede to porting a game to Linux. A company puts out a call, saying "We want to port our stuff to Linux!" and he just answers the call, and dutifully performs the required work in exchange for pay. He's hardly a thought leader; he's hardly someone who grabs the problem of Linux gaming by its horns and leads the masses of corporations into bringing their games to Linux. The company has to make the first move; their CEO or CTO has to say "Yeah, I think that's a good idea... let's put some money behind it".

He's not the uncaring greedy money grabber you portray him as, either. Don't forget the quite large amount of open source projects he spawns during his porting work.

[allquixotic]

He's not the uncaring greedy money grabber you portray him as, either. Don't forget the quite large amount of open source projects he spawns during his porting work.

I don't care how much money he makes; that's irrelevant. He's welcome to all the money that he can (legally ) acquire. And while it is undoubtedly beneficial that he has released these open source tools, there are certain statements and positions that I've heard him make either directly or through Phoronix, which I find repulsive. He, like our friend KameZero, doesn't seem to care about the long-term perpetuation of the core ideals which shaped what Linux is today.

I myself am a pragmatist; I'm not a hard-liner FSF guy, nor am I a hard-liner OSI guy. But I feel that the people who are with us in the long run, as the perpetuators of the free desktop, value either the Free Software or the Open Source movement (or both). People who would just prefer "to get things working" without placing any value on the tenets of either the FSF or the OSI's ideals -- those people just offend me.

So while Icculus does contribute to the perpetuation of FOSS through the release of his open source tools, it's very puzzling to me that he'd turn right around and say that he'd rather that we all use the proprietary drivers so that we don't have to worry about the poor features/performance/whatever of the open source drivers.

The open source drivers are trending toward greatness; it will just take time. There is nothing inherent in the approach of open source graphics drivers that prevents them from being just as good as proprietary ones. In fact, I think they'll end up much better. But in order to resist the temptation of settling with a solution that offends our core beliefs -- that is, the free and open source core beliefs which comprise the very essence of the free desktop -- we have to look past the present day, and think about what is going to be better in the long term.

But in keeping with the tenets of the FOSS movements, we can't just sit idly by and hope that the open source drivers mature by themselves. We have to actively help them reach their goal. We can start by meeting them half-way, by engineering games which use OpenGL APIs that are well-supported by the open drivers; by using semantics that are expected/tolerated by the open drivers; and by (most importantly!) testing our software with the open drivers, and making changes to our code and/or submitting patches upstream whenever we encounter a problem.

These activities, actually, are NOT unique to developing games with the open drivers. As anyone who's worked at a large game studio can tell you, much of the pre-release effort of a game studio is expended trying to get the proprietary driver manufacturers (mainly AMD and Nvidia) to adjust their drivers to allow code interleavings or semantics that are important to the game's engine to run correctly on the binary driver. Even after release, you will often see binary driver updates that say they improve the performance of a certain game by a large percentage (clearly resulting from heavy collaboration between the driver team and the publisher/developer of the game). It seems eminently fair to attempt that same level of collaboration with the open source driver developers.

[allquixotic]

Back on-topic: it'd be good to confirm that they aren't just looking for someone to develop games for Android. Android technically is "Linux", so it's not inconceivable that they could be looking for someone to hack on Valve games for the fastest-growing platform in the world (Android) and its already-huge installed base.

Just looking for clarification to make sure we won't get disappointed: if Valve is just looking to port games to Android, that won't help the Linux desktop at all, because the entire rendering code for Android will be different than what you'd need on an X11/DRI based platform. Also you'd be engineering for "small" embedded hardware, not quad-core desktops with discrete GPUs. Big difference in content quality and the number of quality vs performance shortcuts you take in the renderer.

[xpander]

but icculus is right about opensource drivers. those drivers will play catch up forever.
atm open source drivers are about 5 years behind the prop ones.
yes in some cases for regular desktop use they are great. but 3D perf is terrible.
people whining about why this and that game wont work with opensource drivers.. thats the problem why some devs dont want to release games for linux.
its not really hard to use prop drivers, unless u have prehistoric hardware or some nongaming videocard.


just my 2 cents about that.

[Michael]

Back on-topic: it'd be good to confirm that they aren't just looking for someone to develop games for Android. Android technically is "Linux", so it's not inconceivable that they could be looking for someone to hack on Valve games for the fastest-growing platform in the world (Android) and its already-huge installed base.

Just looking for clarification to make sure we won't get disappointed: if Valve is just looking to port games to Android, that won't help the Linux desktop at all, because the entire rendering code for Android will be different than what you'd need on an X11/DRI based platform. Also you'd be engineering for "small" embedded hardware, not quad-core desktops with discrete GPUs. Big difference in content quality and the number of quality vs performance shortcuts you take in the renderer.

Not unless Android was renamed to Ubuntu....

https://twitter.com/#!/michaellarabel/status/185822756993110018/photo/1

[ownagefool]

I don't care how much money he makes; that's irrelevant. He's welcome to all the money that he can (legally ) acquire. And while it is undoubtedly beneficial that he has released these open source tools, there are certain statements and positions that I've heard him make either directly or through Phoronix, which I find repulsive. He, like our friend KameZero, doesn't seem to care about the long-term perpetuation of the core ideals which shaped what Linux is today.

I myself am a pragmatist; I'm not a hard-liner FSF guy, nor am I a hard-liner OSI guy. But I feel that the people who are with us in the long run, as the perpetuators of the free desktop, value either the Free Software or the Open Source movement (or both). People who would just prefer "to get things working" without placing any value on the tenets of either the FSF or the OSI's ideals -- those people just offend me.

I have to disagree with you there. From your own words you are a fairly hard-liner, border line zealot, or at the very least an idealist. I'm not disagreeing with your position either, given all else is equal I believe OSS > Priority, however a pragmatist accepts that things are not always equal and are much more likely to bend in order to get what they need.

We live in a world where people need to make money, I myself do this my writing software none of which is open source. I don't see any innate right I or anyone else has has over the works of others. Game engines, for example, are typically quite difficult to write, and those who spend the time writing them deserve the option of doing so for monetary reward, and whilst there are other buisness models to achieve this, the most common is to remain closed source, which is exactly what the likes of id software do, at least for a time.

Given all else is equal, of course an open source engine, or any other software, is better as it can be ported and updated as things change. However, as a pragmatist today, I'd rather have steam, my proprietary games, on their proprietary engines than go without, or dual boot windows like currently.

That does not mean I don't have massive respect for the likes of the FSF for trying to work towards a better tomorrow, I just want to get things done today. There is the argument that those of us who bend, are detrimental to achieving the greater good; I'm not sure if I entirely agree with that (a linux with steam is potentially more popular, and thus will garner potentally better support, which with any luck with be OSS) but a guys gotta eat.

Still, respect to you if you use no proprietary software, it's a shame your ideals will require you to miss out on steam for linux if it happens.

[Malizor]

Well, if I was Valve, I would wait for the big upcoming wayland switch before releasing anything (so about a year)?
I mean, we talk about proprietary software here. They may be kind enough to port Source &co to GNU/Linux, but they probably won't want to spend too much time maintaining it.
And Wayland will probably requires invasive changes?

But, anyway, this is a really good news.

[allquixotic]

but icculus is right about opensource drivers. those drivers will play catch up forever.
atm open source drivers are about 5 years behind the prop ones.
yes in some cases for regular desktop use they are great. but 3D perf is terrible.
people whining about why this and that game wont work with opensource drivers.. thats the problem why some devs dont want to release games for linux.
its not really hard to use prop drivers, unless u have prehistoric hardware or some nongaming videocard.


just my 2 cents about that.

So you have a time machine and can travel into the future and tell us that the open source drivers never quite worked out, right? Must be nice to be able to see into the future.

Try plotting two mathematical functions:

y1 = 10x + 100
y2 = 0.1 * x^2
where x = [0, 150]

If you want, go to wolframalpha.com and type in "plot 10x+100, 0.1*x^2 from x=0 to 150".

Let y1 be the proprietary drivers' graph. Let y2 be the open source drivers' graph. Let's say that the y axis represents "overall goodness" (an abstract number taking into account e.g. FPS, driver features, stability, and framerate smoothness).

You're standing at a vantage point of x=10. You can only "see" out to x=15 or so. From this data you're concluding that the open source drivers will "always" be inferior.

If you could only see out to about x=110, you'd see that y2 overtakes y1, and far surpasses it from that point on, dramatically stealing the show.

This is how I see the open source graphics drivers, based on my observations of their progress in the past few years. What evidence are you using to say that they will be behind "forever"?

[kraftman]

Not unless Android was renamed to Ubuntu....

https://twitter.com/#!/michaellarabel/status/185822756993110018/photo/1

What a crap... I mean os x. Those are wonderful news! Bring games to Linux and os x can go home.

[allquixotic]

Well, if I was Valve, I would wait for the big upcoming wayland switch before releasing anything (so about a year)?
I mean, we talk about proprietary software here. They may be kind enough to port Source &co to GNU/Linux, but they probably won't want to spend too much time maintaining it.
And Wayland will probably requires invasive changes?

But, anyway, this is a really good news.

Wayland isn't going to make it to the free desktop until you can start an X11 application within Wayland (either directly within Wayland, or within an X11 server, similar in concept to what OS X does) with full 3d support and 0 regressions. Count on that. They aren't going to start shoving it out in release distros by default until you can take ${FOO_X11_APP} (whether 3d accelerated or not!) and run it with essentially native performance within Wayland.

If they fail at that goal, Wayland will never get off the ground, and we'll continue to use Xorg for the foreseeable future.

Either way, any games that come to Linux/X11 are going to continue to work on future desktops. I can't imagine it being any other way.