Valve's Gabe Newell Talks Linux Steam Client, Source Engine

[yogi_berra]

2. Open sourcing of engine(s) and/or the Steam client itself

Yeah, because I want my credit card numbers being stored in a program where anyone being paid by organized crime can look through the source, find a flaw, and exploit it.

It would be a huge disappointment if they reinvent the graphics stack once again, like Android did and like Chrome OS did, rather than using Xorg.

No, it wouldn't. It wouldn't even be disappointing if X.org disappeared entirely.

[ChemicalBrother]

Yeah, because I want my credit card numbers being stored in a program where anyone being paid by organized crime can look through the source, find a flaw, and exploit it.

You know, that happens with with proprietary software, too. And Valve/Steam got hacked. Closing the source doesn't make it safer. Also opening the source doesn't make it less safe.

[jgardia]

I'm not sure where did you get these numbers from, but they are far off from actuall results. I wasn't talking about Gentoo specifically in my post. I was only suggesting it will cause market fragmentation if only one distro is addressed.

Gentoo and Ubuntu address different markets. Gentoo is designed for power users who can make their own decisions about OS configuration, where Ubuntu's target audience are Linux newbies with no real Linux experience, where decisions are made for them. Obviously Ubuntu will have much bigger market share for that reason. There is plenty of space for both distros on the market.

With all the respect I have for a developer, I have to say that I run Gentoo for about 2 years in my main computer, but I ended up switching to Kubuntu. I do like the concept behind Gentoo, and the amount I learned from that experience allowed me to, among other things, get a better paid and more interesting job.
For me the biggest reason to move was that when a big component of the system was updated (like X or KDE), it took ages to compile. Also, I stated with a very minimalistic set of compile options, and when I discovered that I needed a new option for a specific program, and I had to re-compile a big group of others to have the support, it was really a pain.
Of course, within the things I learned was how to make a good streamlined kernel, and it is something I still do, instead of using the standard ubuntu one. Now that I have a newer machine that is way faster, I'm considering to go back (the feeling was stronger when it was announced that Kubuntu will not be supported anymore), mainly because of the amount of control you have over the distribution.
Anyway, your comment about Ubuntu is for newbies and Gentoo is for power users is a bit pejorative, and even though it might be partially true in some respect, I think you cannot put all the users in the same basket.

[allquixotic]

Yeah, because I want my credit card numbers being stored in a program where anyone being paid by organized crime can look through the source, find a flaw, and exploit it.

Spoken like someone who has no idea how software works. Your credit card numbers are not stored in the Steam client, and nowhere did I ever suggest or mention that I'd want, or that Valve would want, to release the server-side source code to Steam. There is zero security risk to releasing the client-side source code to Steam, unless someone recompiles a version that does steal your identity, releases it, and you're stupid enough to download and use it. But you take this risk any time you download any open source software that is involved in payment transactions, so by your logic, you should only ever run Windows and Internet Explorer, because they're closed source.

It comes down to trust, that's all it is. If you trust the source of the software you're running, that's all that matters. The open or closed source-ness of the software is irrelevant. People still can, and will, discover vulnerabilities even if they don't have the source. And for the best hackers, the reality is that not having the source available doesn't make it particularly harder than it would be if they had it. Especially since most of the vulnerabilities are going to show up in the client-server networking stuff, and that is completely observable with something as common as WireShark.

Besides, someone already hacked Valve and stole a bunch of hashed passwords and email addresses. From their server. Without anything at all to do with the client, and without any source code being released. The Steam client is not doing anything other than attempting to hit the Steam server services, which are already fully accessible on the public Internet for hackers to try and build their own custom client to them, which would "do bad things".

You might also be interested to hear about the extremely convincing, if not downright correct, arguments of a company called Subgraph in favor of open source as an inherent boon to software security. If SSL/TLS, the global standard for secure financial transactions on the web, isn't convincing enough for you -- I don't know what is. The standard is completely in the open; most implementations of the standard are completely open; and yet, attackers would much rather compromise your web browser's javascript implementation or session cookies (things which are unrelated to SSL) rather than try to break SSL itself. If visibility makes exploits easier, then why hasn't TLS 1.0 been cracked?

No, it wouldn't. It wouldn't even be disappointing if X.org disappeared entirely.

Troll much? So you're basically saying that if Valve writes some proprietary graphics stack specifically for their Steambox, and their Linux ports of their games only run on that proprietary graphics stack, you'd be fine with that? Because if we get rid of Xorg, desktop Linux basically has nothing. Zero. Zip. We don't get to use their "Linux" software, period. Just like we can't play Android games on desktop Linux without an extremely slow, software emulator.

I mean, sure, Valve could use Wayland.... but why would they, when it's so immature and under-developed? It would be awesome if they did support it, and it would be a huge impetus to get desktop distros to adopt Wayland; but I'm extremely skeptical that they would even consider it. And even if it did, the people who are stuck with proprietary drivers (hi Southern Islands, hi Kepler) or unbearably slow/buggy open source drivers (hi Evergreen, hi Northern Islands, hi APUs, hi Kepler, hi... oh why do I even bother) would be completely unable to play the games. My bet is that even a full-on adoption of Wayland, by Valve, would not give desktop Linux the ability to run their games. They would either commission or write a proprietary driver for whatever hardware is in their Steambox, and that'd be the end of it.

Oh, also! Just because they might want to port the Source engine to Wayland, or whatever other graphics stack they come up with, doesn't mean that other developers will want to go there too. There are dozens upon dozens of existing games for native Linux, and all of them -- without exception -- use OpenGL, GLX, and... X11. So if that all goes away, you're left with Steambox only running Valve games, and anyone who releases a cool Linux game for X11/GLX in the future is not going to be able to sell it on Steam.

So unless you're totally against Valve bringing their products and content distribution platform to Linux at all, you would be forced to acknowledge that using X11/GLX would be the best way for them to do it. They'd be able to support the best range of existing hardware and drivers (both open and proprietary), the best range of distributions (nearly all), and they could still build a solid and very reliable Steambox based on Xorg. They'd just have to use a proprietary driver, or else find a perfect hardware chipset that's cheap and already has great open driver support, or else, contribute to the open drivers directly to get them to the standard they need for their games. That third option is what I'm hoping for, but will probably not get unless Gabe Newell has been possessed with the combined souls of Linus Torvalds, Brian Paul and Ryan Gordon.

[ownagefool]

Yeah, because I want my credit card numbers being stored in a program where anyone being paid by organized crime can look through the source, find a flaw, and exploit it.



No, it wouldn't. It wouldn't even be disappointing if X.org disappeared entirely.

I'm not sure I understand why you think such functionality would need to exist in the client? An open source client would at best talk to a valve API, likely resulting in alternative UI's, and integration into things such as the ubuntu store. I don't particularly think they'll be open sourcing the client, but it'd still be a cause to celebrate if they did despite the fact that it wouldn't do much by itself.

[cl333r]

I hope Valve doesn't try to bend over backwards to please every weird fringe Linux distro, instead focus only on the few main and modern ones.

[fuzz]

Valve is also hiring more, and it's just not Linux OpenGL -- even if you're just a Linux kernel developer they might be interested in hiring you (I'll be placing some more recommendations to them in the next few days, if you think you might be interested, contact me for possible referral especially if your work has been covered on Phoronix in the past).

Get them to hire some open source driver developers .

[psycho_driver]

I'll still stand by that the original phoronix posts were inaccurate and misleading, and at the very least deserving of apology. Nothing was confirmed back then, none of the so-called release timeframes were proper, and the dead horse was beaten for about 2 years. Even now, all that's shown are some desktops - something I rather suspected Valve to do internally, but still nothing indicative of official support. No actual quotes from Newell himself. Nothing official. UT3 was more official than all this at one point, and look how that turned out.
The point is: until Valve themselves issue a press release, nothing's official. Take everything here with a massive haul pack load of salt. The best you can take from it is that Valve have games running on linux desktops internally within the company. Anything more is pure speculation, and really should be labelled as such within any posts.

You saved me the trouble of having to make almost exactly the same post. Even two years ago there was some truth behind it all since that alpha steam client frontend was found and everybody started playing around with it. However, just because something exists or even if it's being actively worked on doesn't mean it will ever see the light of day.

Nobody should get their hopes up about this until there is official word from Valve.

[Anarchy]

I liked the news so much that I clicked couple of times on the commercials.

[fuzz]

I hope Valve doesn't try to bend over backwards to please every weird fringe Linux distro, instead focus only on the few main and modern ones.

If you can make a .deb, you can package it pretty much anyway you want.

There is this thing called Linux, and in this Linux they have these communities, and in these communities they have helpful people who know what to do....

What's the point if there is no choice?