It's good to see an assumption I had, be validated. The assumption was that Linux would struggle to make a dent in the desktop market until it managed to make a tempting proposition for gamers. Projects like WINE tried, but there's just not enough you can do only with open-source contributors, with such a massive undertaking. I'm happy that a big company is finally backing this effort and with nvidia open-sourcing their gpu kernel modules, I'd say exciting times are up ahead :)
> Projects like WINE tried, but there's just not enough you can do only with open-source contributors, with such a massive undertaking.
Proton is basically just Wine, right? Aside from that CodeWeavers has been doing commercial Wine development for a long time.
Almost all of Wine already existed before Valve got involved. Yes, they help out too, and that's great, but people are hugely over-crediting Valve (and here, hugely under-crediting all those open source contributors, who did the overwhelming majority of the work).
Proton is Wine with some extra patches; the main project has some pretty hefty requirements in terms of coding standards to reduce possible regressions, which means that on the whole it can be pretty slow to get compatibility.
Gaming is just one subsection of the Win32 API that doesn't touch on all parts of said API. Proton basically has a bunch of extra patches that might introduce a regression in upstream Wine but won't in the context of videogames.
Proton patches also usually get upstreamed when they reach "proper" maturity/are fully tested if I'm not mistaken so long-term, Proton benefits the rest of the wine ecosystem too. As for the open source contributors - Valve iirc just hired some of the previously self-employed wine developers and the guy who got I think it was a Vulkan compat layer working on Linux.
I'm pretty sure Codeweavers has the same model although I don't think they upstream patches nearly as much? They're also very much more targeting business customers on Apple devices who need to run some obscure (usually old) Windows program rather than the entire scope of Win32 software.
Basically it's best to look at proton the way you sometimes get a "next generation" fork of a popular piece of FOSS - higher dev speed and more features, however more focused into keeping a single area working than on the overall health of the upstream (which does get contributed back to for the features that help the upstream).
most likely, although by now they also may be employees of valve. but valve is also working with codeweaver.
btw: i consider codeweaver somewhat the unsung hero of the linux desktop.
there is no other company in this world that stakes their success entirely on the success of the linux desktop, or maybe on the demise on the windows desktop, because wine is the only software tool that will never ever be useful running on windows.
> wine is the only software tool that will never ever be useful running on windows
Intel was actually including Wine code in their Arc graphics drivers to translate DirectX 9 calls to Vulkan, because they didn't have performant native DirectX 9 drivers. It's a pretty common trick for Arc users to drop the latest DXVK DLL into the directory of a game that is having issues with the graphics card.
Using Wine DLLs is also a good way to resurrect games built on the obsolete DirectDraw API on modern versions of Windows.
> I'm pretty sure Codeweavers has the same model although I don't think they upstream patches nearly as much?
According to Wikipedia "all changes made to the Wine source code are covered by the LGPL and publicly available", but I don't really know from first-hand experience.
That's the thing with desktop linux: almost everything already exists. But there's always an extra mile to go. Sometimes it's tiny bugs, sometimes it's bad default settings, sometimes it's questionnable UI... all things that may not be a problem for tweakers but will stop people that want a good out-of-the-box experience. I'm glad commercial companies go that extra mile.
i didn't check but i believe the major part of the extra mile is fixing 1000s of tiny little bugs that break games. it should be possible to verify that by looking at the wine changelog. the rest is creating a smooth experience downloading and running games. smoother than any alternative (like playonlinux or lutris)
Wine has worked surprisingly well for a long time, well before Valve got involved, although compatibility was never perfect of course (and still isn't!) I don't think they had to "fix 1000s of tiny little bugs that break games". Codeweavers has been selling commercial Wine-based solutions for a long time.
It's easy to offer a better experience if you also own the actual storefront; that was kind of my point.
And as I said, I'm sure Valve contributes; but describing wine as some sort of semi-usable half-working project before their involvement is rather inaccurate.
describing wine as some sort of semi-usable half-working project before their involvement is rather inaccurate
i disagree. wine worked, but many games didn't. fixing that is mostly valve's doing. and in my opinion the wine experience is still inferior compared to steam.
the problem is very much one of perception. 90% of the work that went into wine was already there. but the remaining 10% of work that was missing make up 90% of the impression of how well it is working. (the numbers are made up, but i hope you get my idea. it's the general linux desktop problem. the desktop is functional and stable, but the impression of it isn't because of a few small issues that just seem to bother many people)
so by attacking these remaining issues, valve is making a significant contribution.
i think it is also important that valve is able to apply patches before they make it into wine, because that means these patches see way more testing than the wine project probably would be capable of on their own. which is another reason that makes vales work significant.
I used wine for many years. "Semi-usable half-working project" was entirely my experience; nearly everything I tried to run had at least one weird bug.
The proton work is open source (and readily installable without using the storefront), so you can actually just look and see that they did, in fact, fix many little bugs that break games - perhaps only many hundreds, rather than thousands.
Have you actually used the commercial version? It's relatively stable and QAed and you can get support from the Wine developers, but it lags the open source version by months. You're not getting the latest compatibility fixes or performance improvements with it.
nevertheless the commercial version must be of use to someone by addressing pain points that other users don't have. and i wager that it is exactly those small issues that don't bother others but that matter for people willing to pay for it.
Theorycrafting here, but wine and proton will make it pretty easy to decouple Steam platform DRM.
It should be noted that it's perfectly possible to release your game on steam in a way where the steam client doesn't even need to be running in order to launch the game. The choice to tie Steam's service into the ability to run the software you purchase /seems/ to be a decision made entirely by the people publishing their games.
If that's the case, you should really be more concerned about that behavior rather than Steam's DRM existing. Don't buy games whose behaviors you disagree with!
Valve is a contributor, so that’s not an entirely true statement. And I was replying to your original statement that “Proton is just wine”, which again, is not true.
Basically means: “in the most essential respects; fundamentally.” So in this case, no, omitting the word basically makes no meaningful change to the meaning of your sentence.
Proton is not just WINE. It is a bundle of many open source technologies including WINE. Proton was a major step forward in the consolidation and simplification of running Windows games on Linux. It also included a ton of upstream contributions to all the bundled technology by full time paid Valve employees.
It's interesting to see Apple basically imitating this with their porting toolkit. The net result is game developers being presented with two growing ecosystems for the same game simply by ensuring they stay compatible with both. With their VR stuff coming out soon, there could be a big push to get some games going there.
Either way, if you are a game developer, ensuring the games work fine via proton and Apple's porting kit should be a no-brainer. Basically, it should just work mostly without too much effort. All you need to do is a bit of Q&A and maybe avoiding to use some stuff that is problematic or coming up with some workarounds for those things. Neither ecosystem is very big. But at the same time, growing your market by a few percent is always nice. It's big enough that not supporting that is becoming an odd choice to make.
I've been using Steam on my Manjaro laptop for some casual gaming. The setup was pretty easy and I'm able to run most of my steam library. I'm not much of a gamer but I've bought some things over the years. And most of those things stopped working on macs. My guess is that Valve is starting to see a lot of Linux users with steam accounts that don't own a Steam Deck. It wouldn't surprise me if that's actually most of their Linux users.
As for OSS, Wine is basically very successful thanks to efforts like this. Most successful OSS projects need active users and, like it or not, the private sector tends to be by far the largest stakeholder in these projects. Wine as a hobby project was basically a bit of a niche thing. They got a lot of amazing stuff done of course. But it was kind of a bit of a pain to set it up and use it and progress was slow. Also support for a lot of games was a bit lacking and very dependent on the community chasing that. Steam makes this super easy and they made huge progress with getting a lot of games working. More importantly, they are big enough that game developers themselves now care about compatibility. And that's of course more commercial users.
Since historically Linux usage hovered around 1%, I'm guessing that the Steam Deck by itself didn't drive any additional users to Desktop Linux (or at least not in huge amounts), though the 1% by itself is probably driven by Valve's efforts to improve Linux support.
It's great that Valve is willing to pay them for it. It's a bit sad that such capital investment from a commercial party is necessary, but I can live with that.
As someone who has just switched to Linux gaming, I've got to say it's working great. Maybe it's slightly less stable, but not by much, and performance is excellent.
It's a bit sad that such capital investment from a commercial party is necessary
by now the linux ecosystem has received to much commercial investment to separate out what is community effort and what is not, but i believe that most of the polish that went into linux desktop applications and interfaces have some commercial investment in them.
the problem is that for most volunteers 90% of usability is good enough. if it works for me, my work is done. on the other hand the imperfect is almost necessary to attract contributors. a perfectly working application is more likely to attract users but less contributors. and as soon as contributors work on something it becomes imperfect again.
this is not to say that contributors mess things up, on the contrary, but most do just enough work to make it work for them, leaving the work to polish it up for other users to someone else. and i think that is perfectly fine. contributors should not be expected to do work that is of no benefit for themselves. the benefit of a polished application is almost always for non-contributing end users, which in turn is almost always a commercial interest.
That's absolutely true. Valve has added significant value and I love them for it. And the community is still doing great work tuning configurations for each game.
One particularly great thing about Valve's support is that it benefits all games, and not just those on Steam. It's absolutely a great example of how open source helps everybody.
> It's a bit sad that such capital investment from a commercial party is necessary
I think it's great that someone invests. That investments are necessary is no sadder than that salaries are necessary for your work. It's just a normal part of work/life where the alternative would be to do it for free, out of passion. But you also have to make a living.
> That investments are necessary is no sadder than that salaries are necessary for your work.
Well, that's also sad.
I just wanted to say it would have been nice if the community would have been able to get this working more easily. But clearly it's too much work to do it without some dedicated fulltimers.
fully agree! I'm biased because I also work for pay on open source products but I think it's a great "rising tide lifts all boats" model of software development.
Well, yes. But what I mean is that it would have been nice if hobbyists were able to accomplish this without needing the help from a commercial entity.
I'm a bit at a loss why some people here prefer commercialised open source over community-driven open source. I mean, I'll take it, and Valve did an amazing job, but there have been plenty of attempts to make this work before Valve stepped in, and it just didn't work that well back then, and it would have been nice if we didn't have to rely on Valve for this.
This sounds the same as you wanting other people to do things for free. I don't know about nice, but I certainly wouldn't lament people being paid to do something.
You're missing my point. The issue isn't money, it's dependency.
Using Steam and the whole Valve-supported infrastructure still doesn't cost me anything. But what if Valve decides the Steam Deck isn't commercially viable for them?
Then we'll be a lot further ahead than if Valve had never paid for this work to be done. I don't know what better alternative you're after. Isn't this open source work? There's no ongoing dependency. If Valve shut up shop tomorrow we're in a better state than if they'd never funded development, aren't we?
it was the opposite for me, I was tempted into getting the steam deck because it seemed like a fun device and an exciting new frontier for linux, and it's gotten me into gaming in a way I haven't been since college.
When I got my SteamDeck, I thought I was going to do a lot more tinkering than I did gaming. I planned on getting NixOS working on it. I ended up just playing games and having fun just using the stock install.
I'm tempted to get one, but it's going to be sitting on a shelf, along with my Oculus Quest. Gaming is a thing that sounds nice in principle, but I rarely can get into it in practice.
i was in the same boat - i wanted to enjoy games again, especially indie games, but i was mostly just buying them during steam sales and then never playing them. i bought the steam deck in order to see if it would be a more fun gaming experience than my laptop, and if not at least i'd have a fun linux device to tinker with. but it absolutely did get me back into playing (maybe an hour or so a day) and i've been enjoying it a lot.
Hmm, that's very interesting. I'm exactly the same as you, I really like indie games but it's always "meh, I'd rather do something else". Maybe I'll borrow a friend's Deck for a weekend to try it, thanks!
That's how reality works. In order to do something, resources must be invested. In order for resources to be invested, people must perceive value in the thing they are investing in.
In order to do something big, many people must invest many resources, and they would only do so if they all see great value in the thing.
In order to have resources to invest, all of the people must have excess resources that they have generated from previous endeavors.
Remember, capitalism is merely the most efficient system we have for assigning limited resources across society.
the interesting thing for me is that my assumption was that linux gaming would not pick up until the desktop experience of installing and using games would improve, but i somehow missed that creating a separate "game desktop" would do the trick. i didn't consider before that it absolutely makes sense to keep games in their own environment which does make the task a bit easier.
but this also validates another assumption of mine, that it would take funding probably driven by commercial interest (and here i was thinking of companies like redhat) to fix this.
> The assumption was that Linux would struggle to make a dent in the desktop market until it managed to make a tempting proposition for gamers.
I don't think that assumption has been validated, and I don't see that from this article. The premise of this article isn't that the steam deck is taking market share, it's that it's increasing compatibility of games on linux. More to the point, the steamdeck hasn't made a dent in the desktop market, see [0] it's a fraction of a percentage of _gamers_ never mind any other use cases.
Don't read this as me saying this isn't progress, but we're still a far cry (pun intended) from a dent.
Linux is now more popular than OSX according to those statistics. That's a decent milestone, and it's only going to grow as enshittification forces further downgrades in user experience, freedom and privacy in commercial OSes.
> Linux is now more popular than OSX according to those statistics.
Linux is more popular than OSX _on steam_ according to those statistics.
> it's only going to grow as enshittification forces further downgrades in user experience, freedom and privacy in commercial OSes.
I think you're wrong here. Take a look at the wayback machine [0] for a history of this data. There was absolutely 0 growth in linux use over the past decade until last year. The thing that changed wasn't enshittifaction, it was a piece of hardware that provided a good user experience. It's nothing to do with windows/mac getting worse or less privacy focused, it's valve providing a smooth usable product that _happens_ to run linux. I'm curious if Proton's success causes a decrease in windows users and an increase in other distros, but right now that's not what we're seeing.
Windows is enshittifying so rapidly that I firmly believe that, if Valve really wanted to, they could eat Microsoft's lunch when it comes to the non-enterprise desktop market, which is almost exclusively gamers these days. Fortunately for Microsoft, Valve seems content to officially position and market SteamOS as just for its handheld, even though it's entirely possible to run it on other hardware.
I am continually in awe of how rapidly terrible its gotten since Windows 7. I loved Win7 - it just worked, with a sleek presentation. Every release after was worse overall (I'm not saying good features weren't added, mind you - but the aggregate experience has been worse with every release for me).
(Note: I fully realize Win7 had a ton of windows warts under the hood - I'm just talking about the user experience)
Ads in the start bar is something that makes me refuse install Windows nowadays. I paid for the thing, why do I get ads? And why do I need an MS account? Just idiotic.
I feel like if there was some way to backport whatever useful things are in windows 11 (and/or windows 8/10/11+ exclusive games somehow detect their expected OS via polyfills) it would be something I might actually pay for.
Or at least someone should upload a no-BS win11 ISO somewhere.
You don't even need SteamOS to use proton. I have played one game on my Windows partition in the last 5 years. Now I assume games will work on my Linux computer until proven differently instead of the other way around.
I don't know. Developing on Windows 11 + wsl has been a suprisingly good dx for me. Much better than my previous MacOS experience. I understand developers who stay in the Apple ecosystem due to the unbeatable laptop hardware in the macbooks, but software wise most of them would be surprised with the current state on the windows side.
I'm very happy with WSL2. The only (well known) problem is the issue with routing when running containers in WSL while behind a VPN. Strangely WSL-vpnkit can't help anymore after some recent updates. It seems WSL 2 2.0.0 will help with this
WSL2 isn’t bad and is getting better but it basically falls apart when you need to run k8s workloads on it. I wish MacOS had native container support - but even without it I’ve not had any issues running a local dev cluster like I did with WSL
Docker with WSL2 runs fine and I use it for my light dev needs on my personal windows machine. Running K8s is the issue.
WSL just had lots of instability issues in my experience when I had to run a handful of services for debugging stuff locally. I tried all sorts of configs to get it to work but it was inconsistent at best. Tons of wasted time. Enough so that our teams got work to get us macs instead (we were a windows shop spinning up a new cloud product - so windows didn’t really make sense as our daily driver anymore anyways). This was about 2 years ago. Maybe things are better now but 6 months of dealing with that instability was enough to sour me.
The issue with gaming on Linux was the typical chicken or egg. Why would companies make games for Linux if no one plays on Linux and why gamers would use Linux if no games are being made for it.
Between wine/proton and more engine OS cross compatibility in general this is starting to become a non issue.
Not that hard anymore to find someone who plays exclusively on Linux and if some game is not there it's not going to buy it.
It is not as much of a failure as just a product released at the wrong time. Most of Steam Deck is basically what lessons Valve has learned with that experiment - Steam Input, Proton, Steam Link (?).
Now that the ecosystem is more mature, there is again some interest towards it. A lot of tech reivew youtubers are now making content on how to get Steam OS to run on X device (X = GPD Win, low profile gaming PC etc..). Honestly, It is good enough to meet most of my "non technical" friends' requirements for their "Home Theater" PC.
> They are using Linux to avoid paying the licences.
Windows already doesn't charge any license fees for screens less than 9" no?
It may have started out that way but I think with projects like DXVK, Gamescope etc.. it is also about controlling the whole software stack to provide a console like experience. Also allows them to do things without worrying about drivers from AMD or some random Windows update breaking their customization.
Another interesting thing maybe Valve's idea of reducing the friction. People used to purchase their games on steam even when pirated copies were easily available.
Now that there are more competitors to Steam (Windows Store, Epic, Uplay, Orign, Gog, etc...), Valve is creating their own vertical to make it easier for people to ignore those places when buying games.
Eg. I had the option of purchasing a DRM free copy of an old game on Gog vs. the same game on Steam Deck. On desktop, I'd have definitely gone with Gog. but on the deck, I actually went with Steam that day. Still baffles me.
On top of the sibling commenter's point (that timing is everything - a 2015 failure doesn't guarantee a 2023 failure), Steam Machine wasn't the same as what the gp is suggesting.
Steam Machine was a PC-like piece of hardware aimed at competing in the console market. The gp is suggesting a gamer-oriented piece of software aimed at competing in the consumer OS market. The latter is definitely not something Steam Machines tried and/or failed at.
Steam Machines had the option of running early versions of SteamOS, which Valve sorta pushed and most manufacturers ignored. SteamOS was pretty rough back then.
It's interesting how this is just like how when Linux started being used in super computers we saw a similar creeping feature improvement for everyone else.
I think that will always be one of the strongest points of Linux, the variety of different areas improvements can come from.
I'm so happy to be wrong about how Proton/Wine taking over might result in a suboptimal user experience for gaming on Linux.
On my Deck, I couldn't care less if the game was a native Linux version or runs via. proton. The experience is so smooth. And more and more game developers are willing to get their games to run on Linux now.
It is honestly nice to see so many famous tech youtubers pushing videos on how to get Steam OS to run on their gaming devices. Thanks to Valve for changing the Gaming Landscape on Linux.
I bought a modest gaming rig in 2016 and used systemd/loginctl to convert it into a multiseat setup. One seat for myself and one for the rest of the family.
The amount of value I got so far from this setup is unbelievable. Part of that is proton taking off and giving me and my kids access to most games on steam.
Very grateful to all involved that made it possible.
Still running ubuntu that have just been upgraded over the years on it.
You will typically do this for a screen card, sound card and some USB ports and then you will have another seat available.
If you want to reset run:
~ sudo loginctl flush-devices
Some gotchas:
Seats and users are two different concepts. You can log in on any seat with any user but then you should not log into the other seats with the same user. That generally causes issues for me but is fixed with a reboot.
Lotta users == lotta bug reports, complaints, and motivated people.
Great example is CP2077, which just got a 2.0 release, and naturally has a lot of issues, with a few that are related to Linux. The one I got was related to not detected the keyboard well via Proton Experimental.
There was a workaround up on steam discussion boards in like < 1 week, and it's getting plenty of attention.
Ha! The hex/exe patch! Had to do the same thing. This issue also sent me down the path of downloading cutter and trying to reverse engineering/interpret the .exe.
> optional case-insensitivity to the F2FS, the Flash-friendly file system
Gosh, I've been lambasted for years for telling Linux folks that case-insensitive file systems are actually good for the average user. Maybe we'll actually get it now that Valve paved the way.
Why would the average gamer on a couch with a steam controller as sole input device (that's what SteamOS is about, isn't it?) care about case sensitivity of file names? It's not like they would actually type those.
Well, Valve is probably adding it to better support games that expect a case-insensitive filesystem. However, case-insensitive filesystems are beneficial for the average business software user. I'm just hoping that they accidentally make this work for ext4 also, and some distro maintainer sees the potential there.
Probably, but I also had to manually convert my home directory to btrfs so maybe this is why I remembered ot wrongly? I seem to remember that earlier steamos releases also had ext4 as rootfs, but I can be wrong.
To be extra sure I downloaded a fresh steamos install and yes my rootfs is btrfs
open source contributions partly funded by corporate time is often a win-win.
i would be cautious about being too optimistic about x year being the "year of linux desktop" just yet tho. but it does indeed have gotten much better if you just play singleplayer games and get all work done on a browser.
