Fragmenting what? There were no solid ecosystem, ever, nor there's any.
There's Linux kernel, GNU base userspace (although some replace it with BSD or Solaris one, huh) and bunch of various software running above those. After this there were nothing common in those software since the very beginning. Just more and less popular software.
Edit: removed X11 as possibly common component, because I totally forgot about DirectFB, huh.
There was a time around 2008 when I thought Ubuntu could be that solid ecosystem. That me, a customer I'm supporting, a developer I'm submitting a bug report to, the author of an FAQ I'm reading, a fellow user I'm helping out in the forums, a package maintainer testing a binary release and a hardware vendor debugging their drivers might all be seeing the same things on our screens and getting the same responses to the same commands at our terminals.
Unfortunately it seems like Canonical does not share my vision.
I can't quite figure that out from the article either, but from the language, I guess they're insinuating that the architecture changes ( in the Ubuntu way as opposed to the traditional "Linux way" ) will mean different approaches to software.
Daniel Stone, contributor to X.Org (one of the base technologies
without which we'd still be staring at blinking green text on a
black background), put in: "I'm not worried about Wayland's
future at all. I'm just irritated that this means more work for
us, more work for upstream developers, more work for toolkits,
and more work for hardware vendors."
I don't get the fuss about Wayland and Mir. Like inventing a new mostly-incompatible subsystem is something new in GNU/Linux world. I mean, there were no "Linux way" ever - just take a look at zoos of sound subsystems, network management subsystems, init/rc subsystems and so on. GNU/Linux world is well-known for having magnitudes of software doing same things but in completely different ways, sometimes with some compatibility layers (like ALSA-OSS), but frequently completely mutually incompatible. This is not good, nor bad thing, it's just that there's no any established rules beyond fundamentals POSIX and FHS (and even those are poked from time to time), just more or less popular approaches.
GTK ran on DirectFB just fine, without X11 stack, so getting rid of X11 is nothing new. There were XDirectFB that worked as a X11 server, too. Not sure if the code's maintained and works now, though - haven't visited that land for, IIRC, about 5 years. Anyway, and I don't remember any worries about that.
I think DirectFB is a bad comparison, since AFAIK it was never used widely by any distro. I think people are right to worry that a new graphics engine will lead to loads of wasted work dealing with compatibility problems and other issues down the road. Based on the massive amounts of problems caused by the sound system zoo, I understand the concern.
Agree. I do like the general idea behind Mir. Canonical is at least trying to get the the "Linux-desktop" thing happening. If that means re-inventing the wheel, then go ahead. If they hit home with this, it will benefit us all Linux users.
No one seemed to care about Gnome fragmenting the Linux Desktop back then, because it had 'the right license'.
It was, however, one of the reasons most big-house commercial software was not ported and sold in Linux versions, and sent Linux into a black hole of license fundamentalists until Ubuntu changed that.
Why this point of view? Because I believe that the best possible computing platform from a technical point of view (the most efficient, better designed, etc) is a combination of both an open source foundation and the possibility of open source and closed source applications on top of it.
There's Linux kernel, GNU base userspace (although some replace it with BSD or Solaris one, huh) and bunch of various software running above those. After this there were nothing common in those software since the very beginning. Just more and less popular software.
Edit: removed X11 as possibly common component, because I totally forgot about DirectFB, huh.