There's going to be an online Linux App Summit this November

The one thing I hope comes out of this summit is a better way forward for application development. Our diversity is one of our greatest strengths and weaknesses (due to fragmentation). I feel like we're often so worried about courting Windows and Mac users and making our software available to their platforms that we've neglected the deep fragmentation in our own.


Last edited by randyl on 11 August 2020 at 4:05 pm UTC

The one thing I hope comes out of this summit is a better way forward for application development. Our diversity is one of our greatest strengths and weaknesses (due to fragmentation). I feel like we're often so worried about courting Windows and Mac users and making our software available to their platforms that we've neglected the deep fragmentation in our own.

I don't really see that big of problem with fragmentation, outside of most commercial vendors seeing it. Like when you ask Adobe to support desktop Linux, they have to ask Qt, GTK, other? Then they have to ask Ubuntu, RH, Debian, Arch...

But the second part of that is pretty easy to answer these days (basically AppImage is the only one so far that doesn't have some people whining about it in the flatpak vs snap debate). 1st one should be chosen on technical / licensing reasons. But let's face it, it's things like the Adobe suite that makes or breaks 'year of the desktop linux', as it is on the list of requirements 'some' people have.

No, the "fragmentation" isn't a bad thing. Think of it in terms of innovations, if only one thing is allowed, will we ever actually get progress?

I mean look at multi-desktop as a good example.
That started as a friendly competition between window managers in the 90's. And then two decades later the first commercial (and inferior ito usability) clone of it appears.

I can't imagine being productive without that feature, for example.

So, keep the fragmentation, keep the ability to choose, because without choice life will be oh so dreary.

As to fragmentation . . . I've never really had a problem installing and running KDE apps in Mate (which is I believe GTK-based). Sure, the initial download of the first one is a bit slow because it brings in a bunch of KDE stuff, but whatever. Maybe people finickier than me would says "But it doesn't look exactly like the Gnome apps!" . . . inquire whether I am concerned.

Recognizing fragmentation issues and solutions doesn't imply all paths lead to homogenization. When we ask a game studio to support "Linux" and provide us a native "port", what exactly are we asking? What are they supposed to target and support? Do we expect them to support all our various packaging systems?

This doesn't even touch on all the Linux ports that only work with specific package systems, distro, or library versions. Most of the titles in consumer gaming fit into the category of closed proprietary binaries that are specifically subject to fragmentation.

No, the "fragmentation" isn't a bad thing. Think of it in terms of innovations, if only one thing is allowed, will we ever actually get progress?

I mean look at multi-desktop as a good example.
That started as a friendly competition between window managers in the 90's. And then two decades later the first commercial (and inferior ito usability) clone of it appears.

I can't imagine being productive without that feature, for example.

So, keep the fragmentation, keep the ability to choose, because without choice life will be oh so dreary.

Yeah, can't tell if Windiws copied Linux fir virtual desktops, or because macOS had them. Also not sure if NeXTSTEP is where those came from or just most other Unix DEs. Either way, from my understanding the Amiga had them first, and independent resolutions to boot!

Recognizing fragmentation issues and solutions doesn't imply all paths lead to homogenization. When we ask a game studio to support "Linux" and provide us a native "port", what exactly are we asking? What are they supposed to target and support? Do we expect them to support all our various packaging systems?

This doesn't even touch on all the Linux ports that only work with specific package systems, distro, or library versions. Most of the titles in consumer gaming fit into the category of closed proprietary binaries that are specifically subject to fragmentation.

In my mind we should be asking 'please support this game on Linux via Steam or GOG or itch.io.' not "please make debian packages."

No, the "fragmentation" isn't a bad thing. Think of it in terms of innovations, if only one thing is allowed, will we ever actually get progress?

In terms of innovation, yes, but no so much in terms of development (I quote below a good illustration):

When we ask a game studio to support "Linux" and provide us a native "port", what exactly are we asking? What are they supposed to target and support? Do we expect them to support all our various packaging systems?

--8<--------------

In my mind we should be asking 'please support this game on Linux via Steam or GOG or itch.io.' not "please make debian packages."

But wouldn't targeting a "store" end up for example like what we see with Epic Game Store getting exclusives ?

The one thing I hope comes out of this summit is a better way forward for application development. Our diversity is one of our greatest strengths and weaknesses (due to fragmentation). I feel like we're often so worried about courting Windows and Mac users and making our software available to their platforms that we've neglected the deep fragmentation in our own.

There is no fragmentation. On Windows every program every secon pwrusr g4m3r uses is a third party program that offers next to no integration to the supposedly "non-fragmented" desktop/toolchain.

As for support, all is really needed is that they figure out a base set of (external) libraries they want/need to support. And steam/gog provides them that already, so even them targeting ancient library versions are not a problem anymore.

To my mind every single dev/publisher that figured out that steam can ship their version of libraries and that it is enough to support one distribution to satisfy 99% of linux gamers has done a good enough job. This has nothing to do with how many desktop environments and how many "linux" distro flavours are out there.

The one thing I hope comes out of this summit is a better way forward for application development. Our diversity is one of our greatest strengths and weaknesses (due to fragmentation). I feel like we're often so worried about courting Windows and Mac users and making our software available to their platforms that we've neglected the deep fragmentation in our own.

There is no fragmentation. On Windows every program every secon pwrusr g4m3r uses is a third party program that offers next to no integration to the supposedly "non-fragmented" desktop/toolchain.

Could you give an example of such third-party program ?

As for support, all is really needed is that they figure out a base set of (external) libraries they want/need to support. And steam/gog provides them that already, so even them targeting ancient library versions are not a problem anymore.

To my mind every single dev/publisher that figured out that steam can ship their version of libraries and that it is enough to support one distribution to satisfy 99% of linux gamers has done a good enough job. This has nothing to do with how many desktop environments and how many "linux" distro flavours are out there.

99% seems a bit optimistic, if we look at the GOL statistics (assuming of course that those statistics are representative)

As for support, all is really needed is that they figure out a base set of (external) libraries they want/need to support. And steam/gog provides them that already, so even them targeting ancient library versions are not a problem anymore.

To my mind every single dev/publisher that figured out that steam can ship their version of libraries and that it is enough to support one distribution to satisfy 99% of linux gamers has done a good enough job. This has nothing to do with how many desktop environments and how many "linux" distro flavours are out there.

99% seems a bit optimistic, if we look at the GOL statistics (assuming of course that those statistics are representative)

Not the point being made. He's not saying nobody uses other distros. He's saying that by and large those other distros, if their users are likely to have the faintest interest in gaming, will ensure that the Steam approach will work on their distro. So game developers code to Steam's de facto standard of what libraries to use and such, distros make sure that works, 99% of Linux gamers find games work. Thus, functionally not fragmentation.
Note that I don't know if this is true, I'm just trying to make sure you're not talking past each other.


Last edited by Purple Library Guy on 12 August 2020 at 6:52 pm UTC

As for support, all is really needed is that they figure out a base set of (external) libraries they want/need to support. And steam/gog provides them that already, so even them targeting ancient library versions are not a problem anymore.

To my mind every single dev/publisher that figured out that steam can ship their version of libraries and that it is enough to support one distribution to satisfy 99% of linux gamers has done a good enough job. This has nothing to do with how many desktop environments and how many "linux" distro flavours are out there.

99% seems a bit optimistic, if we look at the GOL statistics (assuming of course that those statistics are representative)

Not the point being made. He's not saying nobody uses other distros. He's saying that by and large those other distros, if their users are likely to have the faintest interest in gaming, will ensure that the Steam approach will work on their distro. So game developers code to Steam's de facto standard of what libraries to use and such, distros make sure that works, 99% of Linux gamers find games work. Thus, functionally not fragmentation.
Note that I don't know if this is true, I'm just trying to make sure you're not talking past each other.

Thank you for the clarification ; English is not my native langage, so I read it like "[...]it is enough to support one distribution to satisfy 99% of linux gamers"

No, the "fragmentation" isn't a bad thing. Think of it in terms of innovations, if only one thing is allowed, will we ever actually get progress?

In terms of innovation, yes, but no so much in terms of development (I quote below a good illustration):

When we ask a game studio to support "Linux" and provide us a native "port", what exactly are we asking? What are they supposed to target and support? Do we expect them to support all our various packaging systems?

--8<--------------

In my mind we should be asking 'please support this game on Linux via Steam or GOG or itch.io.' not "please make debian packages."

But wouldn't targeting a "store" end up for example like what we see with Epic Game Store getting exclusives ?

Not anymore than 'we sell from our own store, and you can only get this one package that works on *buntu, but you are on your own with anything else."
But otherwise that is already happening. Commercial games for Linux are only really being released on those two, or from individual store fronts. Maybe if we had more choices, like Origin and Uplay, then we could ask for those to be released... as it is, we get mediocre suplort from GOG and only really have Steam and Itch. Unless snap or flatpqk gets some sort of retail support or we get some weird shock like EGS coming to Linux, it is what it is.


Last edited by slaapliedje on 13 August 2020 at 4:08 am UTC

No, the "fragmentation" isn't a bad thing. Think of it in terms of innovations, if only one thing is allowed, will we ever actually get progress?

In terms of innovation, yes, but no so much in terms of development (I quote below a good illustration):

When we ask a game studio to support "Linux" and provide us a native "port", what exactly are we asking? What are they supposed to target and support? Do we expect them to support all our various packaging systems?

--8<--------------

In my mind we should be asking 'please support this game on Linux via Steam or GOG or itch.io.' not "please make debian packages."

But wouldn't targeting a "store" end up for example like what we see with Epic Game Store getting exclusives ?

Not anymore than 'we sell from our own store, and you can only get this one package that works on *buntu, but you are on your own with anything else."
But otherwise that is already happening. Commercial games for Linux are only really being released on those two, or from individual store fronts. Maybe if we had more choices, like Origin and Uplay, then we could ask for those to be released... as it is, we get mediocre suplort from GOG and only really have Steam and Itch. Unless snap or flatpqk gets some sort of retail support or we get some weird shock like EGS coming to Linux, it is what it is.

I've found Itch to be a bad experience overall because the games target old OS versions that are either statically linked to out of date libraries or dynamically linked in a manner that breaks when libs change. Dynamically linking sounds smart but every library maintainer has their own idea of how backwards compatibility works. But don't worry, fragmentation isn't a problem.

Has anyone ever asked why GoG doesn't support us? Don't ask too deeply or you won't like the answer. Market share always takes the blame, but it isn't the only reason. We're a huge pain in the ass because "supporting Linux" means everything and nothing at once. It isn't worth their effort.

Why won't EGS "come to Linux"? Because it isn't worth the effort for a fraction of a fraction of marketshare for a distribution that doesn't even know if it wants to be a desktop OS or not. Canonical can't even figure out a plan for 32 bit blob support going forward. They just backtracked on that a little and pushed it out to be dealt with some other time.

There is no fragmentation or consistency issue. This is fine.

No, the "fragmentation" isn't a bad thing. Think of it in terms of innovations, if only one thing is allowed, will we ever actually get progress?

In terms of innovation, yes, but no so much in terms of development (I quote below a good illustration):

When we ask a game studio to support "Linux" and provide us a native "port", what exactly are we asking? What are they supposed to target and support? Do we expect them to support all our various packaging systems?

--8<--------------

In my mind we should be asking 'please support this game on Linux via Steam or GOG or itch.io.' not "please make debian packages."

But wouldn't targeting a "store" end up for example like what we see with Epic Game Store getting exclusives ?

Not anymore than 'we sell from our own store, and you can only get this one package that works on *buntu, but you are on your own with anything else."
But otherwise that is already happening. Commercial games for Linux are only really being released on those two, or from individual store fronts. Maybe if we had more choices, like Origin and Uplay, then we could ask for those to be released... as it is, we get mediocre suplort from GOG and only really have Steam and Itch. Unless snap or flatpqk gets some sort of retail support or we get some weird shock like EGS coming to Linux, it is what it is.

I've found Itch to be a bad experience overall because the games target old OS versions that are either statically linked to out of date libraries or dynamically linked in a manner that breaks when libs change. Dynamically linking sounds smart but every library maintainer has their own idea of how backwards compatibility works. But don't worry, fragmentation isn't a problem.

Has anyone ever asked why GoG doesn't support us? Don't ask too deeply or you won't like the answer. Market share always takes the blame, but it isn't the only reason. We're a huge pain in the ass because "supporting Linux" means everything and nothing at once. It isn't worth their effort.

Why won't EGS "come to Linux"? Because it isn't worth the effort for a fraction of a fraction of marketshare for a distribution that doesn't even know if it wants to be a desktop OS or not. Canonical can't even figure out a plan for 32 bit blob support going forward. They just backtracked on that a little and pushed it out to be dealt with some other time.

There is no fragmentation or consistency issue. This is fine.

Mountain of molehill...
You know why there are so many different Linux distributions? Because different people have different needs. You don't think that if Windows was open source and 100% customizable that there wouldn't be all sorts of crazy and awesome configurations? You don't think the same would be said about iOS? Choice is good. There isn't so much as fragmentation as there is a lack of understanding how to package things cleanly. A platform like Steam does it mostly correctly with few bugs. AppImage is another method. Flatpak as well. These are things that actually are trying to solve the packaging / library differences you're so concerned with.

If you're preferred use of Linux is to just play games on it, use GamerOS or one of the other ones that are optimized as such. Me, I enjoy general computing usage and gaming, so I use Debian. The fragmentation thing is really a myth, as any distribution that doesn't suck will have work arounds to make them more usable. Look at Arch, generally not the friendliest to install, but you can keep the same install updated for many years, and the user community is fantastic and make all sorts of PKGBUILDs for just about any software you can think of. Does this make them different than Debian, Ubuntu, RHEL/Cent, Fedora? Sure! Is it a good or bad thing? I think it's a good thing, as I sad, we all have our use cases!

Are you a developer for anything Linux related? If not, I'd say talk to some and find the real issue. Sure support can kind of be a pain from the "well I have Arch installed and...' well when did you last update? "three months ago..." "Update, then see if the issue is still there..." But I mean if you're running Arch... you should be punched if you did that.

It's simple enough to say 'we'll help with whatever issue, but if it takes more time because you're not running 'supported' OS than we may have to give up, as we can't support every config.' That's basically the 'correct' way to support software on Linux.

No, the "fragmentation" isn't a bad thing. Think of it in terms of innovations, if only one thing is allowed, will we ever actually get progress?

In terms of innovation, yes, but no so much in terms of development (I quote below a good illustration):

When we ask a game studio to support "Linux" and provide us a native "port", what exactly are we asking? What are they supposed to target and support? Do we expect them to support all our various packaging systems?

--8<--------------

In my mind we should be asking 'please support this game on Linux via Steam or GOG or itch.io.' not "please make debian packages."

But wouldn't targeting a "store" end up for example like what we see with Epic Game Store getting exclusives ?

Not anymore than 'we sell from our own store, and you can only get this one package that works on *buntu, but you are on your own with anything else."
But otherwise that is already happening. Commercial games for Linux are only really being released on those two, or from individual store fronts. Maybe if we had more choices, like Origin and Uplay, then we could ask for those to be released... as it is, we get mediocre suplort from GOG and only really have Steam and Itch. Unless snap or flatpqk gets some sort of retail support or we get some weird shock like EGS coming to Linux, it is what it is.

I've found Itch to be a bad experience overall because the games target old OS versions that are either statically linked to out of date libraries or dynamically linked in a manner that breaks when libs change. Dynamically linking sounds smart but every library maintainer has their own idea of how backwards compatibility works. But don't worry, fragmentation isn't a problem.

Has anyone ever asked why GoG doesn't support us? Don't ask too deeply or you won't like the answer. Market share always takes the blame, but it isn't the only reason. We're a huge pain in the ass because "supporting Linux" means everything and nothing at once. It isn't worth their effort.

Why won't EGS "come to Linux"? Because it isn't worth the effort for a fraction of a fraction of marketshare for a distribution that doesn't even know if it wants to be a desktop OS or not. Canonical can't even figure out a plan for 32 bit blob support going forward. They just backtracked on that a little and pushed it out to be dealt with some other time.

There is no fragmentation or consistency issue. This is fine.

Mountain of molehill...
You know why there are so many different Linux distributions? Because different people have different needs. You don't think that if Windows was open source and 100% customizable that there wouldn't be all sorts of crazy and awesome configurations? You don't think the same would be said about iOS? Choice is good. There isn't so much as fragmentation as there is a lack of understanding how to package things cleanly. A platform like Steam does it mostly correctly with few bugs. AppImage is another method. Flatpak as well. These are things that actually are trying to solve the packaging / library differences you're so concerned with.

If you're preferred use of Linux is to just play games on it, use GamerOS or one of the other ones that are optimized as such. Me, I enjoy general computing usage and gaming, so I use Debian. The fragmentation thing is really a myth, as any distribution that doesn't suck will have work arounds to make them more usable. Look at Arch, generally not the friendliest to install, but you can keep the same install updated for many years, and the user community is fantastic and make all sorts of PKGBUILDs for just about any software you can think of. Does this make them different than Debian, Ubuntu, RHEL/Cent, Fedora? Sure! Is it a good or bad thing? I think it's a good thing, as I sad, we all have our use cases!

Are you a developer for anything Linux related? If not, I'd say talk to some and find the real issue. Sure support can kind of be a pain from the "well I have Arch installed and...' well when did you last update? "three months ago..." "Update, then see if the issue is still there..." But I mean if you're running Arch... you should be punched if you did that.

It's simple enough to say 'we'll help with whatever issue, but if it takes more time because you're not running 'supported' OS than we may have to give up, as we can't support every config.' That's basically the 'correct' way to support software on Linux.

Thanks for that perspective. I'll take it into consideration.

The one thing I hope comes out of this summit is a better way forward for application development. Our diversity is one of our greatest strengths and weaknesses (due to fragmentation). I feel like we're often so worried about courting Windows and Mac users and making our software available to their platforms that we've neglected the deep fragmentation in our own.

There is no fragmentation. On Windows every program every secon pwrusr g4m3r uses is a third party program that offers next to no integration to the supposedly "non-fragmented" desktop/toolchain.

Could you give an example of such third-party program ?

As for support, all is really needed is that they figure out a base set of (external) libraries they want/need to support. And steam/gog provides them that already, so even them targeting ancient library versions are not a problem anymore.

To my mind every single dev/publisher that figured out that steam can ship their version of libraries and that it is enough to support one distribution to satisfy 99% of linux gamers has done a good enough job. This has nothing to do with how many desktop environments and how many "linux" distro flavours are out there.

99% seems a bit optimistic, if we look at the GOL statistics (assuming of course that those statistics are representative)

An example? Sure, every non-windows program, including the drivers. All the stuff that used to come on CDs. Then the stuff to overclock your hardware, the countless programs to clean windows, antivirus or browsers. (though i guess people just use edge and windows defender now) I don't understand how it's less fragmented than what we have. You are not forced to run a kde/gnome/xfce/openbox frankendesktop, you can just as well run one of them and pretend the others don't exist, same as on windows.

For games all of this stuff is largely unimportant, since they will only use a handful of libs like sdl to interact with your window manager/DE of choice. and any version mismatch can be bridged by the way steam and gog handle these things. It won't eliminate all possible errors under the sun but that's avoided by the existing approach to pick one officially supported distro for a game.

An example? Sure, every non-windows program, including the drivers. All the stuff that used to come on CDs. Then the stuff to overclock your hardware, the countless programs to clean windows, antivirus or browsers. (though i guess people just use edge and windows defender now) I don't understand how it's less fragmented than what we have.

Yet your initial statement was that were no fragmentation (emphasis mine):

There is no fragmentation. On Windows every program every secon pwrusr g4m3r uses is a third party program that offers next to no integration to the supposedly "non-fragmented" desktop/toolchain.

The difference being, a program for Windows targets a version, not a flavour. Which means that it will work, no matter if you are running Windows <insert version here> Home / Pro / Business. On Linux, packages exists for some distributions only (if any), and after then it is up to the user to sort things to make it work.

You are not forced to run a kde/gnome/xfce/openbox frankendesktop, you can just as well run one of them and pretend the others don't exist, same as on windows.

I have seen plenty of threads on forums where someone asked how to do this and that, and usually, the first answer was along the lines of : "which DE are you using ?" or "I don't know for <insert DE #1>, but on <insert DE #2>, do [...]"
Which usually end up on some obscure command line (so modern... ) which may also target a certain distro (aka "sudo apt-get [..]" which is royally ignored on Manjaro and stuff...)

For games all of this stuff is largely unimportant, since they will only use a handful of libs like sdl to interact with your window manager/DE of choice. and any version mismatch can be bridged by the way steam and gog handle these things. It won't eliminate all possible errors under the sun but that's avoided by the existing approach to pick one officially supported distro for a game.

This is exactly the point: developers end up resorting to targeting one (or a few) distro, and it's up to the users to check how to install missing stuff on their specific distro (aka "not out-of-the-box experience"). They cannot simply target, say, Linux 5.4 (or whatever is the latest stable kernel version) like they target Windows 10. Also, package names tend to change between distros, so the user has to double-check the package name in his distro.

I think you may be confusing the issues here.
There is distinctly different "fragmentations" being spoken of here:

1) Fragmentation of the administative environment. Debian/RH/Gentoo/Arch bases all administrate very differently. This fragmentation is being felt, but RPM/APT and Portage/Packman solve different problems. So we actually need much of this.

2) Fragmentation of the app environment. Other than having to load styles seperately for GTK/Qt, we don't really care here. An app built for Gnome/Unity just works in a KDE enviroment.

3) Fragmentation of APIs. This has been a much bigger issue in Windows (DLL hell) than in Linux, and we have a few interfaces that are stable (kernel-userspace + libc) and you can just bundle the libraries on there, and problem solved.


One claims that fragmentation-3 is a non issue, and the other claims that is false because of example fragmentation-1.

This is about an app summit, so they are mostly concerned with fragmentation 2 & 3, and not 1.
If you solve 3, then it will run on any distribution (and run on future ones), and if you solve 2 then the experience will be consistent on any desktop.

I think you may be confusing the issues here.
There is distinctly different "fragmentations" being spoken of here:

1) Fragmentation of the administative environment. Debian/RH/Gentoo/Arch bases all administrate very differently. This fragmentation is being felt, but RPM/APT and Portage/Packman solve different problems. So we actually need much of this.

2) Fragmentation of the app environment. Other than having to load styles seperately for GTK/Qt, we don't really care here. An app built for Gnome/Unity just works in a KDE enviroment.

3) Fragmentation of APIs. This has been a much bigger issue in Windows (DLL hell) than in Linux, and we have a few interfaces that are stable (kernel-userspace + libc) and you can just bundle the libraries on there, and problem solved.


One claims that fragmentation-3 is a non issue, and the other claims that is false because of example fragmentation-1.

This is about an app summit, so they are mostly concerned with fragmentation 2 & 3, and not 1.
If you solve 3, then it will run on any distribution (and run on future ones), and if you solve 2 then the experience will be consistent on any desktop.

Maybe you don't use Windows but 3 is not fragmentation and isn't really an issue at all. Multiple DLL versions (e.g. Microsoft VS C++ runtimes, .net version libraries, etc) can live side by side.

There is no real fragmentation in the Windows ecosystem because it targets a consistent framework and they put backwards compatibility before all else, even when it appears stupid and clunky. Even now with IE11 being deprecated for future development it will be supported and backwards compatible going forwards for the foreseeable future.