Linux Thread

[madrang]

With Steam now available on Mac OSX, the GNU/Linux community is now eager to see a native client. This thread is for those of us who wish to express our desire for such a client and to discuss GNU/Linux gaming in general.

TROLLS BE WARNED: We will not hesitate to report any of this nonsense on sight, and we have had the support of the forum staff for this thread for over a year. Do not waste your time here.

First, let's take care of some myths...

Myth: Steam will never come to Linux!
Fact: Steam for Linux has already been announced, along with a native port of Left 4 Dead 2 and more games to come! You can keep track of Valve's progress on the Valve Linux Blog.

Myth: Nobody uses Linux!
Fact: Millions of people use it, it's just not the several hundred million users that Windows has.

Myth: There are no games for Linux!
Fact: flibitijibibo has taken the liberty of creating a wiki to disprove this in the case of Steam games. Check out The Big List of Steam Games on GNU/Linux!

Myth: Nobody plays games with Linux!
Fact: We have not only proven that we care about games with projects like Wine, but we have also demonstrated that we buy more games/pay more for games than the typical Windows/OSX user in every single Humble Indie Bundle.

Myth: Supporting Linux isn't worth it!
Fact: Not only has the Humble Bundle shown otherwise, but other developers have seen this too. Check this post by the developers of Osmos, this post by the developer of Mystic Mine, and this post by the developers of World of Goo.

Myth: Supporting all the Linux distributions is too hard!
Fact: We have POSIX for a reason. While packaging tends to suck (.deb, .rpm, etc.), this largely does not affect games. Games can be installed with a simple and (mostly) universal .run binary file that installs the game's binaries and content in a single folder, and the necessary libraries (such as OpenAL) can typically be installed from the distribution's repositories. This has been seen primarily with id Software's games, as well as several games in the Humble Indie Bundles.

In fact, even a .run file would be largely irrelevant for Steam games. Steam itself could be installed with a .run and the games would simply be downloaded just like they do now. It's the equivalent of .tar.gz packed games, which are compatible with all maintained distributions.

As for actually developing the games on the platform, it's not any harder than supporting the Mac. In fact, if you use some of the newer features of OpenGL (such as tessellation in OpenGL 4), supporting Linux is actually easier than supporting OSX.

Myth: Linux hardware support is garbage!
Fact: As of 2011 this is now limited to wireless drivers and more esoteric, outdated hardware. Even graphics card drivers are now up to speed, even if they are generally proprietary software.

Myth: Wine sucks!
Fact: Wine... well, yeah, it kind of does. That's why we would like native support, see?

Myth: You shouldn't use Linux for games, just use Windows like everyone else.
Fact: Hey, wait, that's not a myth! That's just a bad suggestion!

Not only do GNU/Linux users just prefer the OS over Windows, but in some cases Windows is just too expensive, particularly if you do not live in the United States. We feel that we should not be forced to purchase an operating system to play games, especially when we could be putting that money towards more games...

So! With that now out of the way, here are some links and resources...

We want Linux client!, a Steam group for GNU/Linux gamers.
The Big List of Steam Games on GNU/Linux, a wiki that lists native games as well as lists of games compatible with Wine according to...
The Wine AppDB, the official database for Wine program compatibility.
The Humble Indie Bundle, probably the biggest public support for Linux games there is right now.
Desura, a client for distribution of games (like Steam).

Thanks to...

flying sheep, for Steam4Linux Answers
flibitijibibo, for The Big List of Steam Games on GNU/Linux

[airblad3]

Hopefully Valve will give us an official Linux release in the near future.

[Midnitte1]

Originally Posted by airblad3

Hopefully Valve will give us an official Linux release in the near future.

Hope so... then we can demand an android version!

[flibitijibibo]

I'll be brutally honest. In attempting to clean up the current OP, I had to stop not just because of all the mechanical errors, but because it A: does not cover everything, B: is too difficult to read, and C: wastes LOTS of space pointing out things that most people already know (security is the big one). While citing articles is good, there are just way too many and people will get distracted, stop reading, and either just leave or post hatespam. The facts are much shorter than the quotes.

I tried merging the new stuff in it with my old one, here's what I got:

EDIT: The OP is now what was here, neat! Thanks madrang!

Cut it from here to prevent wall o' text.

[Sir_Brizz]

I actually like the Myth/Fact format a lot.

[leahcim]

Originally Posted by Sir_Brizz

I actually like the Myth/Fact format a lot.

Well yeah, except it's generally "one opinion / a different opinion" rather than a list of facts.

We know that Valve have linux knocking around (for the source dedicated server) We know that Valve pretty much let people scratch whatever itch they think will give customers benefit (from lots of interviews and final hours of portal 2 etc)

We also know they didn't so much write a native opengl source as a native directx->opengl thing to translate source (which one of their programmers talked a bit about in Mac thread and how he thought performance wise it was on par with a native port - in a thread on here somewhere)

Which suggests to me they either have no one interested in doing it, or they've already done it for a laugh, and there are compelling reasons why they've not decided to release it.

As for me, I've written at length in the past on the topic. I'm not a huge fan of the idea for several reasons

- Linux is great because you get the source. That's the reason I (once, many years ago) joined the folk working on dri and started bug fixing the radeon 32 driver. It was fun. Pretty much every linux game and utility crashed or had rendering bugs, but it was fun finding and fixing them and trying to get quake 3, RTCW and so on running. More fun than playing the games if I'm honest.

But it was going nowhere. Then nvidia came out with their closed source driver and everyone installed that.

I think the same with this idea for Steam on linux - it would be great fun to write. But mostly pointless. (To me this is what wine is too, something that is fun for the developers to play with, but completely useless and pointless)

As the OP points out, once you start playing games on linux, you may as well be in windows (or mac if you want to use unix). Because everything becomes proprietary - the drivers, the games and so on. So it makes the whole thing pointless imo.

I'd much prefer Valve concentrate on good windows (and mac if they must) support for their games and service because the only things I use linux on I either don't care about the OS (my phone, my TV, my router) or the whole idea is that I have the source code.

- I think the games that have been ported to linux have always been done poorly. If the games themselves don't lack features or content (which happens), it's usually true that stuff like mod tools, editors and so on are dropped.

Worse, they get someone in the community to hack together some half-baked, statically linked port that no one in his right mind would connect to the internet (but I bet you all do) So what? Well, I think the point is that most of the list of games people give are bad examples. If Valve do this I'd want Valve quality and attention to detail. The same kind of thing they did for Mac. That makes this more expensive to do (in terms of time and resource) than a quick port - and even the quick port isn't cheap.

- Lastly, I buy lots of games from Steam, including but not limited to Valve's games. Unless a significant %age of those games are on linux too, I can't use linux even if I wanted to.

Not having the games I play and the games in the future I want to play would make it a mere novelty. The truth is, people making lists of games that run on linux are kidding themselves a little bit when it's clear that the list is bereft of thousands of popular games. It's like living in the old USSR and pretending there are cars because, "look, there's a trabent or a yugo and John Carmacksi has got this French guy to sent us the bits to build a Ferrari, minus a few bits and pieces, mine crashed into the wall first time I used it"

[zebcarnell]

Originally Posted by leahcim

- Lastly, I buy lots of games from Steam, including but not limited to Valve's games. Unless a significant %age of those games are on linux too, I can't use linux even if I wanted to.

Not having the games I play and the games in the future I want to play would make it a mere novelty. The truth is, people making lists of games that run on linux are kidding themselves a little bit when it's clear that the list is bereft of thousands of popular games. It's like living in the old USSR and pretending there are cars because, "look, there's a trabent or a yugo and John Carmacksi has got this French guy to sent us the bits to build a Ferrari, minus a few bits and pieces, mine crashed into the wall first time I used it"

That might be an issue you personally feel is a problem. But for the future of PC gaming and for the future of Linux, a Steam and Source build would go a long way. After Microsoft turned their back's on PC gaming (prioritising Xbox and quiting the PCGA) the PC gaming industry needs someone to step up to the plate. I personally never want to be stuck in the hardware restrictions of a console, I never want to be stuck in the software restrictions of Windows and I never want to be stuck with Apple's hardware/software overheads.

[leahcim]

Originally Posted by zebcarnell

I never want to be stuck in the software restrictions of Windows

Heh, this is why it's difficult to take zealots seriously. Think about the subject of this thread. You are stuck, that's why you're here, but not in windows, in linux.

Valve produce software for multiple platforms. They aren't anti-MS, or anti-console, or anti-windows or anti-mac. They might have technical or business reasons why they like some of the features on one platform over another. But, they aren't your saviours. You won't get Valve to produce a linux client by appealing to them on those grounds.

Besides, as I've already said, it makes no sense to talk about open source or any of that (all of which, in context, makes some sense) when you're talking about commercial software using 3D on Linux. It's nearly all closed source (or, unfortunately, crap)

In fact, this is writ large where the idea of open source fails because the output of Valve is very specialised - you can see another sub-thread in this thread has gone off on a tangent with another area linux suffers because it needs exceptional talent or specialised skills to produce the goods - and those people want to eat and either IBM aren't willing to fund them (because IBM want to eat too)

Copying stuff from specs? Sure. Getting the comp sci grads to produce the mind numbingly tedious stuff like standard libraries or network protocols? Not a big problem.

But creativity, art or niche areas? Forget it. If it exists, it's generally commercial. Lots of Hollywood famously uses Linux, but they aren't Open source and they have their own software to do the stuff that linux hasn't yet demonstrated an ability or interest in doing.

Let me explain. There is a linux TF2 server. So, all you need to do is write a client for it. You want openness. You want hippy idealism then write your own client. Then you won't be restricted by Valve or anyone else. In fact, before you write it, write 100 words about why the wonderful linux community hasn't already done it if it's such a big itch that needs scratching.

I suggest that if it's not lack of skill, then it must be lack of interest.

When you've finished that work on your Steam client next.

All software platforms are restricted. Some have numerous long threads begging companies to write or port software to them (hint)

[zebcarnell]

Originally Posted by leahcim

Heh, this is why it's difficult to take zealots seriously. Think about the subject of this thread. You are stuck, that's why you're here, but not in windows, in linux.

Valve produce software for multiple platforms. They aren't anti-MS, or anti-console, or anti-windows or anti-mac. They might have technical or business reasons why they like some of the features on one platform over another. But, they aren't your saviours. You won't get Valve to produce a linux client by appealing to them on those grounds.

Besides, as I've already said, it makes no sense to talk about open source or any of that (all of which, in context, makes some sense) when you're talking about commercial software using 3D on Linux. It's nearly all closed source (or, unfortunately, crap)

In fact, this is writ large where the idea of open source fails because the output of Valve is very specialised - you can see another sub-thread in this thread has gone off on a tangent with another area linux suffers because it needs exceptional talent or specialised skills to produce the goods - and those people want to eat and either IBM aren't willing to fund them (because IBM want to eat too)

Copying stuff from specs? Sure. Getting the comp sci grads to produce the mind numbingly tedious stuff like standard libraries or network protocols? Not a big problem.

But creativity, art or niche areas? Forget it. If it exists, it's generally commercial. Lots of Hollywood famously uses Linux, but they aren't Open source and they have their own software to do the stuff that linux hasn't yet demonstrated an ability or interest in doing.

Let me explain. There is a linux TF2 server. So, all you need to do is write a client for it. You want openness. You want hippy idealism then write your own client. Then you won't be restricted by Valve or anyone else. In fact, before you write it, write 100 words about why the wonderful linux community hasn't already done it if it's such a big itch that needs scratching.

I suggest that if it's not lack of skill, then it must be lack of interest.

When you've finished that work on your Steam client next.

All software platforms are restricted. Some have numerous long threads begging companies to write or port software to them (hint)

Its about Freedom of choice so when Valve say things like this in an interview "I mean, philosophically, our goal is that you buy a game and then you can access it on whatever machine you’re on. So that was the idea with the Steam play concept of if you already own this on the PC on Steam and there’s a Mac version out, you can play it on a Mac. Just go get a Mac or go to your brother’s house or whatever and play it on the Mac."
I expect that to be the case ("buy a game and then you can access it on whatever machine you’re on") My choices and reasons are completely my own and if you don't like my "hippy idealism" then that's entirely your opinion, no amount of reasoning will change my mind. At the end of the day it was a rant, I don't even take my own rants seriously but it seems some people do.....

Originally Posted by leahcim

I think the same with this idea for Steam on linux - it would be great fun to write. But mostly pointless. (To me this is what wine is too, something that is fun for the developers to play with, but completely useless and pointless)

As the OP points out, once you start playing games on linux, you may as well be in windows (or mac if you want to use unix). Because everything becomes proprietary - the drivers, the games and so on. So it makes the whole thing pointless imo.

There are people like me (participating in Blender development BTW), who hate it to reboot their machine just to play some game, if you need some recreational break time.

Linux is my default working and media environment. I do all my coding there, I listen to music and watch videos with on that Linux box, I arrange audio using Linux software (Ardour, Rosegarden). But only to play games I have to reboot into Windows.

But in Windows I don't have access to my Linux partitions (they're in JFS) and much of my regular stuff for work isn't there either.

Having OSS NVidia GPU drivers would be cool, but honestly I don't consider that such a huge problem.

DRI has it's own issues and IMHO needs some major ovehaul – Wayland is a huge step backwards, I mentioned several of the issues of its device and rendering model to the developers, but I got the impression they didn't understand (expect ♥♥♥♥ty rendered fonts to be an issue again, especially in multidisplay setups, messed up color management, poor user input performance in high CPU load situations).

And then there are the crappy sound systems. OSS3: totally outdated. OSS4: can't do MIDI and has no power management. ALSA: Totally screwed up, an unfixable mess. PulseAudio: Trades audio quality for perfomance (introduces huge roundoff errors at each routing step that involves sample format conversion, even if the converson goes to more bits per sample).

I totally love Linux (the kernel) and what I can do with it (X11/GLX OpenGL forwarding, VirtualGL), but the general state of the media infrastructure is a bit pitty.

Nevertheless: The situation is actually good enough for games: The NVidia propriatary drivers offer excellent performance and doing sound using OSS (either native OSS or via ALSA's in kernel compatiblity layer) delivers low latency and high quality at low CPU loads. I'd not bother with PulseAudio, it may be the current trend but I think it will be obsoleted by some well founded, written from ground up sound system sooner than later.

[Temar]

Originally Posted by heXAriTH

PulseAudio: Trades audio quality for perfomance (introduces huge roundoff errors at each routing step that involves sample format conversion, even if the converson goes to more bits per sample).

Interesting, it's the first time I hear someone complaining about roundoff errors in PulseAudio. I tried to find some more information with Google but couldn't find anything useful. Do you have some links where the problem is discussed?

Originally Posted by Temar

Interesting, it's the first time I hear someone complaining about roundoff errors in PulseAudio. I tried to find some more information with Google but couldn't find anything useful. Do you have some links where the problem is discussed?

So far only the bugtracker issue I opened exists. In January I was auditing PulseAudio code and found several issues with the way PA deals with sample conversion. I reported this to the developers. This is an exceprt from the reply I got from the project lead (german original and translation):

Irgendwie glaube ich nicht, daß es Wert ist das zu fixen, zumal nen Fix
vermutlich mehr CPU-Zeit kostet. Wäre nicht die einzige Stelle, wo wir
niedrige CPU-Load oder Cache-Pressure gegen mathematische Korrektheit
traden. Machen wir etwa bei Lautstärkeanpassungen, bei der wir
SW-Anpassung unterdrücken wenn die HW-Anpassung "gut genug" passt
(d.h. unter nem Threshold falsch ist), und wir uns so sparen können
jedes Sample berühren zu müssen.

PA is keine mathematische Beispiel-Rechnung und auch kein
professionelles Audio-Produktions-Tool sondern nen Real-Life
Sound-Server für Consumer-Anwendungen.

Somehow I don't think it's worth fixing that, for such a
fix probably took more CPU time. That wouldn't be the only place
we're trading CPU load or cache pressure against mathematical
correctness. We do that on volume adjustments, for which we
inhibit SW adjustment if the HW adjustment works "well enough",
(i.e. is wrong below some margin) and thus can save touching each
sample.

PA is not some example mathematical calculation and neither a
professional audio production tool, but a real life sound server
for consumer applications.

EDIT/UPDATE:
I just took a look at the ticket I opened then: It's still open and it seems some developers actually do care.

[Mercymizery]

Ok i see where this is going with the latter posts im going to agree with him on the state of media in linux for audio hardware He is right it really is a mess. Ok Like god i hate to use the word windows we have one universal video render open gl. What linux community needs to do honestly i know im comparing it to windows even tho i despise windows its something it kinda did right. The game pad and Audio subsystem need to be standardised and handled completely by the os. Look around at some fourms getting game pads to work in quit a few games is a mess for those that enjoy those. ALso Regardless of distribution there needs to be one set standard audio subsystem Granted even in windows u have eax or no eax as part of an advanced feature set but the os well the direct x api does handle the audio portion quit well. Hell dont reinvent the wheel just rip it off. Is linux viable for gameing atm absolutely just a few tweaks here and there with getting some developers on board. People will flock to it in droves considering the jarate poor state of windows 8 sorry i really wanna know what microsoft was thinking then they came out with that really...

Ps what i mean by standard is. NO matter what distrubiton or if ur kde or gnome we need a audio and controller subsystem that works and works well. N not alesa and or oss for this pulse audio for that.

[flibitijibibo]

Just got told that the OP has been updated. And it is!

Check out the new OP if you haven't already and feel free to rip it apart.

Edit: Also, just for giggles, I contact Humble Bundle Inc. to see if they have any stats for who redeemed Steam/Desura keys. I'm very curious to see if there's any notable percentage of Linux users that grabbed those keys.

[RobbieThe1st]

While I'll be happy with a bog-standard x86-compatible Linux Steam version, it seems like these days code should be made using platform-independant/processor-independant code, so that, for example, the Linux Steam client can be compiled for x86, Amd64 and ARM all without any modifications - Whenever a new platform comes out, it'd be nice to see an (unsupported) version for it.
That's the one thing that OSS has over most all other platforms currently - I can get a version of Debian for ARM or what-have-you that's functionally identical to the x86 or Amd64 versions, and allows me to run any and all OSS software on it.

Now, if we /did/ have a few different versions of a Steam Linux client for ARM etc, it'd be trivial to make them run on Android, Maemo, MeeGo or even iOS(with a lot more support libraries): It'd simply require someone to build a wrapper(and install)-script for the OS in question, and it'd probably work. In some cases, it'd need extra bits, but that's not a huge problem. Games, well... you'd have to find ones that were platform-compatible. Again, it'd be nice if Valve had some sort of Unified Build Archetecture, so all games they have the source to could be compiled for almost all platforms.

Too bad that'll /never/ happen... Oh well.