Denial doesn't mean [LWN.net]

Denial doesn't mean

Posted Dec 23, 2015 5:35 UTC (Wed) by rqosa (subscriber, #24136)
In reply to: Denial doesn't mean by khim
Parent article: Android on the desktop: Not really “good,” but better than you’d think (Ars Technica)

> After that productivity apps and games must be ported.

No need to port them — instead, just make a single compositor that can run both SurfaceFlinger clients and Wayland clients (including XWayland).

> And then, finally,

… you'll be able to run any Android app on any Linux distribution. Then the only remaining difference between "Android" and "a Linux distribution" will be that "Android" has the proprietary Google stuff preinstalled.

> support for Ubuntu and other Linux distributions will be dropped.

If that's true, then how is it that pretty much all of the "productivity apps and games" in question have already been ported to Windows (running on Qt / GTK+ / wxWidgets for Windows, and/or Cygwin, etc.), and Mac OS X too (I remember some people saying that it would kill desktop Linux back when it was new, too), but despite that, desktop Linux distributions are still developed and used today?

to post comments

Denial doesn't mean

Posted Dec 23, 2015 8:46 UTC (Wed) by khim (subscriber, #9252) [Link] (2 responses)

No need to port them — instead, just make a single compositor that can run both SurfaceFlinger clients and Wayland clients (including XWayland).

Who would do that? And, more importantly, why?

If that's true, then how is it that pretty much all of the "productivity apps and games" in question have already been ported to Windows (running on Qt / GTK+ / wxWidgets for Windows, and/or Cygwin, etc.), and Mac OS X too (I remember some people saying that it would kill desktop Linux back when it was new, too), but despite that, desktop Linux distributions are still developed and used today?

You live in some imaginary world. Or may be in parallel universe. In our universe most “productivity apps and games” are only supported on Windows and Linux support is an afterthought at best. Linux desktop survives on outskirts of technology landscape using some leftovers from server platforms where Linux thrives, but since Android would use the same kernel and you would be able to run the same stack of server technologies you run today on Linux desktop… who would need GNU/Linux? And why?

P.S. Note: I'm not saying that Linux distributions will die. They will just become mostly irrelevant. Whoever cares about new version of AmigaOS or KolibriOS? No, really? It's clear that such people exist (or new versions of these OSes wouldn't be released), but how many of these are out there?

Denial doesn't mean

Posted Dec 23, 2015 12:13 UTC (Wed) by rqosa (subscriber, #24136) [Link] (1 responses)

> And, more importantly, why?

The same reasons why any of these exist: Wine, XWayland, various X servers for Windows (and the one for Mac OS X, and there's even at least one Android now), WinUAE and Amiga Forever, etc.

> In our universe most “productivity apps and games” are only supported on Windows

But you said: "productivity apps and games must be ported [to Android]. And then [emphasis mine], finally, support for Ubuntu and other Linux distributions will be dropped."

I thought this was supposed to mean that porting Linux-desktop apps to the future Android desktop will cause the decline of traditional Linux distributions… but apparently you actually meant to say that porting Windows-only apps to Android is what will cause that? How exactly would that hurt Linux usage share, when the apps in question already don't run on Linux?

> Linux desktop survives on outskirts of technology landscape using some leftovers from server platforms where Linux thrives

Not really, it's more like the Linux desktop survives on the outskirts of the desktop-hardware landscape that's designed to run Windows (previously DOS)… or rather that's how it used to be, but even today many of the traditional GNU/Linux distributions support ARM hardware (e.g. ARM Chromebooks, DragonBoard and the rest of the 96Boards family, Raspberry Pi, etc.) And that trend will only increase, especially if/when "Android for the desktop" arrives.

Or to put it another way: Today, Android is the dominant mobile platform, and CyanogenMod survives on its hardware base. In a future where "Android for the desktop" has displaced Windows as the dominant desktop platform, all the "traditional Linux" / "GNU/Linux" / "Unix-like Linux" distributions will have adopted the Android-desktop-hardware base as their own (in the same way that they currently use the Windows hardware base, and as CyanogenMod uses the current Android mobile hardware base), turning CyanogenMod into basically just another desktop distribution. (They'll even be able to grab whichever pieces of AOSP / CyanogenMod / whatever else is necessary to be able to run unmodified Android apps natively… and even will be able to, upon installation on a stock Android device, preserve the preinstalled Android apps from a stock Android device in the same way that CyanogenMod already can.)

> you would be able to run [on Android] the same stack of server technologies you run today on Linux desktop…

The same thing is basically true for both Mac OS X and Windows, even though they don't use a Linux-derived kernel.

(I seem to remember reading an opinion piece circa 2002, maybe written by Tim O'Reilly, that basically said that all Linux desktop users should switch over to Mac OS X, since basically all open source programs for Unix-and-X can be compiled for it. Clearly, that hasn't happened.)

> They will just become mostly irrelevant.

But they have always been "mostly irrelevant" to the majority of desktop users; this majority only uses Windows. The current desktop-Linux user base uses it despite the fact that pretty much all free software for Unix can (one way or another) be run on Windows too. (And it being slightly easier to run traditional-Unix apps on Android, compared to running them on Windows or Mac OS X, isn't going to make much of a difference to that user base.) Likewise, CyanogenMod is irrelevant to the majority of mobile users, and the current CyanogenMod user base uses it despite the fact that they don't need to use it instead of stock Android.

So, I predict that if/when "Android for the desktop" takes over, the majority of desktop users will be using stock Android, whereas today's desktop-Linux user base will then be using the same kind of desktop hardware as the majority (just like they're currently using x86_64 hardware built for Windows) but will have replaced its preinstalled stock-Android OS with some sort of "mutant GNU/AOSP distribution" (in the same way that people run CyanogenMod on Android-mobile devices today)… and the current crop of GNU/Linux distributions (Arch, Debian, Fedora, etc.) will eventually become those mutant GNU/AOSP distributions.

TL;DR version: We are the GNU/Linux user/developer base. Your OS's hardware platform and APIs will be added to our own. Resistance is futile.

Denial doesn't mean

Posted Dec 23, 2015 16:37 UTC (Wed) by raven667 (subscriber, #5198) [Link]

I don't think you are disagreeing so much on the facts or even in predictions of the future, the difference is in the question of relevancy, for the Free Software Linux Desktop to be relevant it would need a lot closer to the installed base of Mac OS X, the non-Free UNIX Desktop. With relevancy you get at least some of the resources (money and time) of all desktop software developers in the industry, not just Free Software fans, with that attention you get the ability to set de-facto standards for software development across the industry and the resources to build really nice things. You can find many ways in which the mere existence of Mac OS X in sufficient numbers has changed expectations from 20 years ago when Windows was dominant, or look at how the existence of Firefox changed the landscape from when IE was dominant. In addition having a large number of users makes the work building software more meaningful, it's not only about helping oneself, it's about making life better for your human community, changing the world and all that comes with it.

On the other hand, while having relevant market share has plenty of benefits, for a software project to be sustainable it really only needs a relatively small critical mass of developers and users, development environments like COBOL or TCL or LISP are still maintained today even though their relevancy was decades ago because there are still enough people who care, today GNOME and KDE have developers and put out releases even though they are in no danger of being relevant to the wider software industry any time soon. For the few million (out of billions) people who use them that may be fine, as long as they have their own needs met, it's not important that others needs are met as well. It's the reason people are so critical about the design direction of GNOME 3, in an attempt to meet the needs of the billions they seem to have de-prioritized the needs of the the few million or thousand technical users who made up the bulk of the user base a decade ago.

I can say from my personal experience in IT that all of the IT staff where I work who were using Linux desktops switched to Mac OS X about a decade ago with 10.4 and haven't looked back. I still am interested and toy around with Linux desktops personally but haven't switched back to Linux for my daily driver although I did just switch from iOS to Android for my mobile devices and use a lot of apps from F-Droid because I do prefer Free Software but it's not my most important criteria.