Steam Linux Use Dips Back Below 0.5%

[Weasel]

http://www.improbability.net/loki/

You can run quite a few programs from the 1999 on current day Linux distributions without recompiling them you do need to provide a run-time.

https://en.wikipedia.org/wiki/Loki_Entertainment Loki existed 1998 to 2001 and people still run those games with work around on current day Linux distributions.

Of course these runtimes will be able to be made better in future.
https://meetings-archive.debian.net/...ing-the-p.webm
The above work.

Yeah, now compare that to the amount of old games Wine can run and see the difference.

https://wiki.winehq.org/Copy_Protection

Wine project is willing to support copy protection and anti-cheat stuff that can be supported. Only one I know that is on the not to be supported list is PunkBuster that thing is check summing the dll files. So unless you have Microsoft dll files the check sums will not match. Yes PunkBuster has at times banned people for running Windows 10 insider edition so its even anti to those running windows at times.

LOL, you don't even READ what you link and you say you're only aware of PunkBuster? Look at the Copy protection that doesn't work" list. I can tell you for sure that ALL korean anti-cheat bullshit (GameGuard, Xigncode, etc), AND copy protection crap like Denuvo and sometimes Themida, don't work and will likely never work because they need drivers or other hacks they rely on.

Note that they will NEVER work on Linux, even without Wine, so it's not Wine's fault at all. Those games will NEVER have Linux ports unless they take out the anti cheat or copy protection.

An example is Deus Ex Mankind Divided: they took out Denuvo on the Linux port, because it simply would never run otherwise. So shut the fuck up you don't know anything.

LOL no moron this is wrong.

https://www.evenbalance.com/pbsetup.php

Above is what annoys the wine project a lot. You have native Linux punkbuster for native Linux programs that works perfectly. Yet when running under wine you cannot have punkbuster. So there is a lot of copy protection and anti-cheat stuff that works on native Linux programs that you cannot have running a windows program under wine.

You have an obsession with punkbuster I guess, but I got a lot more trouble tracking shitty bugs due to other packers and anti cheat, with no way to fix (yes, unlike you, I contribute to it).

From the general ways you post so much clueless crap in general about how programming works, I know you haven't ever submitted a single patch anywhere, so please stop sparing me "how Wine works", I'm aware of it far more than you are, given that I actually try to fix bugs in there. So shut up, you're not an authority on Wine, you're just a clueless parrot.

[Weasel]

This is exactly what I"m talking about, you take one of the best things about linux and use it as a complaint... wtf.... I'm a gentoo user so I'm gonna generalize for gentoo of course, but the fact is as long as I have the right sources going down the whole dependency tree including build time and runtime dependencies, I can in fact recompile anything that was once known to compile correctly. Gentoo is quite old and there are ebuilds that can be reused or modified. It may not be a simple project, but it certainly can be done. So yeah, this is something you could call "behavior not consistent with Windows". And that is an attribute that can be ascribed to linux libraries and tools and programs of all sorts. Many times there are no direct equivalent behaviors...

Sorry to break this to you, but Linux does not enforce open source apps or programs. Just because you're happy with recompiling everything doesn't mean others are, you know, people who want to use closed source software (like Adobe Flash when it was relevant). So it's a valid criticism.

Which brings me to a damn good point... Runtime behavior is emergent..... It would take billions of unit tests to sample expected runtime behavior across every single windows application that exists. The fact is that wines unit tests don't cover even a tiny proportion of behavior that can and does occur.

They don't have to, though. What matters is that the majority of applications don't rely on corner cases.

If a corner case behavior does not match, does it really matter if no application relies on it? That's the whole point behind Wine, since it will never catch up to Windows unless Microsoft open source it.

[oiaohm]

Which brings me to a damn good point... Runtime behavior is emergent..... It would take billions of unit tests to sample expected runtime behavior across every single windows application that exists. The fact is that wines unit tests don't cover even a tiny proportion of behavior that can and does occur.

I believe the wine developers have themselves caught up in a Banach-Tarski paradox.

Not really you have missed the The Zipf Mystery or the 80/20 rule. 80 percent of applications use less than 20 percent of the API this true of Windows/Linux/OS X/Android..... This number goes down into each individual function so 80%*80% gives you about 60% so you could in theory run about half the applications implementing 20 percent of the ABI/ABI and only testing 20 percent of those functions.

Just because behaviour can happen does not mean any application in fact depends on it.

Please note wine own test suite finds that Windows own ABI behaviours are not 100 percent stable either. This is even better for wine since this means application developers could not depend on those functions to work right so compacts the list of what has to be implemented correctly.

So out of the 20 percent of the API/ABI 80 percent of applications use under windows there is a good percentage the applications don't expect to work right anyhow. Now wine problem is filling in that 20 percent of the ones applications really expect to work and stubbing those applications don't care about.

[oiaohm]

Yeah, now compare that to the amount of old games Wine can run and see the difference.

But that was not what you said. You said about running old Linux games. Loki was the closed source dominate Linux game provider 1998 to 2001.

How many Linux applications from 1999 can you run on today's distros without recompiling them?
The answer is over 95% of them if you are willing to give them a run-time.

LOL, you don't even READ what you link and you say you're only aware of PunkBuster? Look at the Copy protection that doesn't work" list. I can tell you for sure that ALL korean anti-cheat bullshit (GameGuard, Xigncode, etc), AND copy protection crap like Denuvo and sometimes Themida, don't work and will likely never work because they need drivers or other hacks they rely on.

https://github.com/wine-mirror/wine/blob/master/programs/winedevice/device.c

Wine currently loads drivers as normal user program level. Lot of copy protection that does work under wine does in fact use drivers. Of course wine driver support would need to be expanded to use like qemu for drivers that need to see ring 0. There has been prototype attempts in past using qemu this way on some of the really picky copyprotections/anti-cheat.

Some programs with Denuvo do work under wine.

https://bugs.winehq.org/show_bug.cgi?id=40631#c16

Denuvo is one of the ones that Denuvo itself does not seam anti-wine. Its more game produces themselves setting different options in Denuvo resulting in finding wine defects or setting the options that tight that the game even fails with Windows updates.

An example is Deus Ex Mankind Divided: they took out Denuvo on the Linux port, because it simply would never run otherwise.

Really this is you being clueless. Lot of the copy protections have equals on Linux.

Themida has Code Virtualizer https://www.oreans.com/products.php

Denuvo does not come in a Linux form that is why Deus Ex Mankind Divided had to remove it for Linux port. They did put something else in it place.

Note that they will NEVER work on Linux, even without Wine, so it's not Wine's fault at all. Those games will NEVER have Linux ports unless they take out the anti cheat or copy protection.

This is kind of wrong. Yes there are a lot of programs to port to Linux that will have to change their anti-cheat/copy protection system. Not they need to be without those systems.

Please note Windows PunkBuster is the only one on that list that I know is impossible to get working its method is confirmed as cannot be fixed.

Items like GameGuard, Xigncode and Themida the research into them is not done. Denuvo is on the list as known to be able to got working but needing lots more work to get every Denuvo using game that can be working.

The list of known impossible is 1 and that is punkbuster for windows. List of unknown is quite a few. The list of 100% possible needing work is quite a few as well.

I will give you 12 months ago Denuvo was in the unknown pile with a lot of people suspecting it was impossible that turns out not to be the case.

[Weasel]

But that was not what you said. You said about running old Linux games. Loki was the closed source dominate Linux game provider 1998 to 2001.

How many Linux applications from 1999 can you run on today's distros without recompiling them?
The answer is over 95% of them if you are willing to give them a run-time.

I'm not sure what your point is? Of course they're going to run if you provide them the runtime. That's... common sense. I'm asking out of the box. Because that's what people complain to Wine about: they don't want to fiddle with it.

If you supply the runtime, the only requirement is that the kernel's interface is stable, because that is literally the only difference you have compared to a distro from 1999: the kernel. So yes it will work, because the kernel is solid and designed well.

https://github.com/wine-mirror/wine/...evice/device.c
Wine currently loads drivers as normal user program level. Lot of copy protection that does work under wine does in fact use drivers. Of course wine driver support would need to be expanded to use like qemu for drivers that need to see ring 0. There has been prototype attempts in past using qemu this way on some of the really picky copyprotections/anti-cheat.

Some programs with Denuvo do work under wine.

https://bugs.winehq.org/show_bug.cgi?id=40631#c16

Denuvo is one of the ones that Denuvo itself does not seam anti-wine. Its more game produces themselves setting different options in Denuvo resulting in finding wine defects or setting the options that tight that the game even fails with Windows updates.

At this point, I don't know what you're trying to argue for.

The point was that many of these "breakage" in Wine isn't due to Wine being bad or having bugs, and I'm sick of people complaining about them as if it's somehow Wine's fault. That's the whole point. It's due to shitty copy protection/anti cheat mechanisms that would never work on Linux the way they are. If Wine virtualizes drivers in the future that only makes it even better that people should stop whining about it: you would NEVER get such thing from native binaries' copy protection.

Denuvo does not come in a Linux form that is why Deus Ex Mankind Divided had to remove it for Linux port. They did put something else in it place.

Thanks for proving my point? People cry that Wine can't run an app with Denuvo for example, and then they don't realize that Denuvo simply doesn't work on Linux. It's not a Wine problem. What the fuck are you even arguing for then? You just proved my point or what?

Please note Windows PunkBuster is the only one on that list that I know is impossible to get working its method is confirmed as cannot be fixed.

Items like GameGuard, Xigncode and Themida the research into them is not done. Denuvo is on the list as known to be able to got working but needing lots more work to get every Denuvo using game that can be working.

The list of known impossible is 1 and that is punkbuster for windows. List of unknown is quite a few. The list of 100% possible needing work is quite a few as well.

Sorry man. GameGuard and plenty of others are literally impossible at this stage.

Note that some wish-thinking like a ring 0 virtualizer can also work with PunkBuster *in theory* since it would have full control over it. But it's not gonna happen, realistically speaking.

And anyway, they would certainly never run on Linux without Wine. in this case, Wine should be praised for even trying to make them work, since native would never run them the way they are.

Nobody will fucking write drivers for an unstable Linux kernel (the driver interface is unstable) just for anti cheat. Nobody.

So you can understand, when people are bashing a project that's dear to my heart for at least trying to make stuff work that would never see a port in its current state, they really really annoy me. (if you ask for them to remove copy protection for the Linux port, might as well ask them to do the same on their Windows port btw, and then it will work in Wine).

[Weasel]

And well, I'm not going to argue further on this, as it's pointless and I'm kind of annoyed right now (not at you oiaohm, just in general, people just can't appreciate the effort Wine puts into compatibility).

But I'm not bashing the Linux kernel either. I fucking just hate intrusive anti cheat/copy protection mechanisms. Already argued with starshipeleven in another thread about it. The fact is, they are hostile to LINUX, not to Wine. Even for native ports, they'd have to be removed or tweaked or replaced with something less intrusive if they would ever run on Linux.

Wine happens to run on Linux so of course, so obviously they break on Wine. It's not Linux's fault at all, I also hate Windows for allowing this cancer to exist and would really love it when Microsoft get fed up with all the nasty problems they bring even on Windows (like BSODs) and say "that's it, we're blocking such practice". Then they'll have no choice but to drop this garbage, and will start to run on Wine/Linux too.

[duby229]

Sorry to break this to you, but Linux does not enforce open source apps or programs. Just because you're happy with recompiling everything doesn't mean others are, you know, people who want to use closed source software (like Adobe Flash when it was relevant). So it's a valid criticism.

They don't have to, though. What matters is that the majority of applications don't rely on corner cases.

If a corner case behavior does not match, does it really matter if no application relies on it? That's the whole point behind Wine, since it will never catch up to Windows unless Microsoft open source it.

You don't have to be sorry to me, I'm not new, sorry to break it to you, but that's -the- reason why many people don't use windows.

And of course wine doesn't know, they don't have the shear number of unit tests they would need to know. My point anyway was that's the reason they are wrong.

EDIT: Oh and btw, I like how you call the vast majority of windows runtime behavior corner cases.... nice.... I could give any example and you would claim corner case...

[oiaohm]

I'm not sure what your point is? Of course they're going to run if you provide them the runtime. That's... common sense. I'm asking out of the box. Because that's what people complain to Wine about: they don't want to fiddle with it.

If you supply the runtime, the only requirement is that the kernel's interface is stable, because that is literally the only difference you have compared to a distro from 1999: the kernel. So yes it will work, because the kernel is solid and designed well.

Please note 1999 is too old for application from windows to work under Windows 10 without major hacking around. Runtime could be a chroot.

Once you get past a particular point out of the box on windows does not work either. Reality here is with effort Linux you can go back a very long way and programs work as good as ever did..

IThe point was that many of these "breakage" in Wine isn't due to Wine being bad or having bugs, and I'm sick of people complaining about them as if it's somehow Wine's fault. That's the whole point. It's due to shitty copy protection/anti cheat mechanisms that would never work on Linux the way they are. If Wine virtualizes drivers in the future that only makes it even better that people should stop whining about it: you would NEVER get such thing from native binaries' copy protection.

Clueless wonder. High end high cost CAD for chip design that run on LInux. Have copy projections just as bad as games.

Thanks for proving my point? People cry that Wine can't run an app with Denuvo for example, and then they don't realize that Denuvo simply doesn't work on Linux. It's not a Wine problem. What the fuck are you even arguing for then? You just proved my point or what?

Sorry man. GameGuard and plenty of others are literally impossible at this stage.

There is a difference between currently don't work and impossible. There are plenty that don't work that we don't know if they are impossible or not to make work.

Note that some wish-thinking like a ring 0 virtualizer can also work with PunkBuster *in theory* since it would have full control over it. But it's not gonna happen, realistically speaking.

There is a will not fix bug for punkbuster. Yes Punkbuster was tried with ring 0 virtualizer. Even with full control PunkBuster not going to work. You have to generate a value from check-summing the binary and uploading it to the server. Yes have a different salt value each time based on the current expect game state. To get the correct checksum you have to have the correct dlls that punkbuster knows and scan memory of the active game. Correct dlls require a legal copy of windows. At this point you are kind of toast. Punkbusters design there is no possibility ever to work full around it without owning a copy of windows. So its truly without question impossible unless punkbuster decides to support wine.

INobody will fucking write drivers for an unstable Linux kernel (the driver interface is unstable) just for anti cheat. Nobody.

DKMS drivers you find with closed source cads and closed source flight simulators for Linux they give you list of supported Linux distributions and if you decide to run your own kernel good luck to you.

Please note some of these drivers are for their input redirecting evdev to their program. Everything you see windows based anti-cheat based stuff do except this if for cad and flight simulators. We have been fairly luck that this stuff is priced too high for most parties releasing games on Linux being willing to pay. But if the number of gamers on Linux increased a lot this could change. Basically the same nasty copy protection/anti cheat stuff does exist for Linux.

Linux kernel is starting to introduce Symbol namespaces.

https://lwn.net/ml/linux-kernel/[email protected]/

Symbol namespaces will make it simpler to know what functions will not be changing in the Linux kernel.

The Linux kernel driver interface is a mixture of unstable and stable. Just has lacked nice way of seeing what one is what. Nvidia closed source drivers using the same binary blob on multi Linux kernels with wrapper code going to known stable functions.

The biggest barrier is not going to be kernel symbols and the like its going to be UEFI secure boot. By the time Linux games could have enough market to afford the horrible version of anti-cheat on Linux you are going to need the drivers signed by the distribution for them to work without major hacking around.

ISo you can understand, when people are bashing a project that's dear to my heart for at least trying to make stuff work that would never see a port in its current state, they really really annoy me. (if you ask for them to remove copy protection for the Linux port, might as well ask them to do the same on their Windows port btw, and then it will work in Wine).

Problem is you see a lot of games have not portable anti-cheat but works under wine then get ported to Linux and then they use punk buster. So if they do the same change to windows version then Windows version would not work under wine. So we really do need to find out what ones are possible to be made work as if they change may make more problems.

Copy protection and anti cheat don't go away just because game comes to LInux normally they change to a cross platform one some of those cross platform ones are still quite bad in other ways..

[Weasel]

You don't have to be sorry to me, I'm not new, sorry to break it to you, but that's -the- reason why many people don't use windows.

Compiling from source? Dude, even amongst Linux users... Gentoo users are a tiny minority.

EDIT: Oh and btw, I like how you call the vast majority of windows runtime behavior corner cases.... nice.... I could give any example and you would claim corner case...

"Vast majority" and yet the majority of Windows apps (that have APIs implemented I mean) work just fine under Wine...? Obviously apps that don't have APIs implemented are not gonna work, that has nothing to do with corner cases, just not implemented yet (e.g. D3D12).

Please note 1999 is too old for application from windows to work under Windows 10 without major hacking around. Runtime could be a chroot.

chroot is the entire distribution minus the kernel. That's exactly what I said. Seriously.

Clueless wonder. High end high cost CAD for chip design that run on LInux. Have copy projections just as bad as games.

Unless they require root access, then no, not "as bad as games". Your judging ability makes no sense in this context. If they do require root, then well, let's just say... those are ahem "specialized" applications that probably only support a single distro (probably some version of RHEL). Dunno, might want to give some examples so I can look them up.

Yes Punkbuster was tried with ring 0 virtualizer.

Source?

Even with full control PunkBuster not going to work. You have to generate a value from check-summing the binary and uploading it to the server. Yes have a different salt value each time based on the current expect game state. To get the correct checksum you have to have the correct dlls that punkbuster knows and scan memory of the active game. Correct dlls require a legal copy of windows. At this point you are kind of toast. Punkbusters design there is no possibility ever to work full around it without owning a copy of windows. So its truly without question impossible unless punkbuster decides to support wine.

I don't think you understand the fact that the kernel (or a ring 0 virtualizer) can control a user process (punkbuster). This has nothing to do with Wine, cause I'm not arguing about Wine anymore. It could redirect the files to fake virtual data for example, so that the checksum matches.

I don't care about PunkBuster. My main gripe is with anti-cheat that require admin privileges (or root on Linux). It's a disgusting practice even on Windows. I wish Microsoft would say "enough" and stop it dead in its tracks. Because clearly that's one thing that will never run on Linux, as people don't run arbitrary software as root.

[oiaohm]

chroot is the entire distribution minus the kernel. That's exactly what I said. Seriously.

The thing is that works.

Unless they require root access, then no, not "as bad as games". Your judging ability makes no sense in this context. If they do require root, then well, let's just say... those are ahem "specialized" applications that probably only support a single distro (probably some version of RHEL). Dunno, might want to give some examples so I can look them up.

TSMC CAD for silicon design is one. Yes it protection does require root access in fact in kernel module that use to decode silicon node into memory and protect it from coping unencrypted..

Source?I don't think you understand the fact that the kernel (or a ring 0 virtualizer) can control a user process (punkbuster). This has nothing to do with Wine, cause I'm not arguing about Wine anymore. It could redirect the files to fake virtual data for example, so that the checksum matches.

Did you miss the bit about salt that they are doing salt based on current state of game. As in what functions of the libraries are in fact loaded in memory and checksum of those functions.

So you cannot just feed punkbuster fake data. Punkbuster would require fake data that has to be aligned with the current actions in the game. So this is going to overhead kill you.

I don't care about PunkBuster. My main gripe is with anti-cheat that require admin privileges (or root on Linux). It's a disgusting practice even on Windows. I wish Microsoft would say "enough" and stop it dead in its tracks. Because clearly that's one thing that will never run on Linux, as people don't run arbitrary software as root.

High end copy protection systems for CAD for silicon designers under Linux has custom encrypted file access yes this requires higher than root privillage..

The worst I have seen requires using custom hypervisor. When it a case you have to use the silicon companies cad or you will not get your custom silicon made they kind of have you over a barrel. Stuff they get away doing as demands makes game companies seam good. Like having to spend 200 dollars to replace a dongle that they have not had made quality so falls apart after 3 inserts.

Linux people don't like running arbitrary software as root but some software you don't have a choice and worse it software you spend over 10000 dollars a seat. We don't need game companies going the Software protection dongle route. Then saying if you don't want to spend a fortune on dongle replacements because we made them crap you will run our program as root or worse inside our custom hypervisor with the distribution of our choice(yes lot of these cads don't provide Windows versions either its their chosen Linux or nothing). CAD copy protection makes game copy protection look angelic.

We are lucky that most of the sane priced closed source on Linux if it has licensing control is either steam or FlexNet Publisher that are both kind of sane. As soon as the software crosses 2000 dollar a seat you start seeing root level and custom kernel driver requiring. When software crosses 10000 dollars a seat you start seeing the really nasty stuff like custom hyper visor enforced copy protection that gets worst like listing what motherboard/cpu/video card/harddrive brand.... you most have so their software works.

I don't think anyone playing games would put up with hypervisor enforce copyright protection. Hypervisor enforced copy protection makes general gaming copy protection and anti-cheat look tame.