So, I got into Linux almost 15 years ago now. Back in the day, Linux was about having a cheap and convenient UNIX-like system for those who couldn’t afford expensive Sun hardware at home or who wanted to get more involved with understanding and hacking on the internals. Linux benefited enormously from the fallout of the “UNIX Wars”, which had seen different UNIX vendors attacking one another, huge amounts of fragmentation etc., before a re-unification effort centered around common standards, like POSIX (and later, SUS). Yes, these standards are not perfect, but they send a strong message of intent.
Because we had common standards for low-level pieces, for interoperability, etc. we could at least make an effort to have portable software between very different systems. The reality wasn’t always rosy, but the intent was there. Everything from networked filesystems (NFS) to graphical desktop (X) was centered around understanding that there were “others” out there you needed to play nicely with. You could go implement some shiny new feature in a silo, or make your version do things in a radically new way, but the implosion of the UNIX market had demonstrated the futility of doing that in a vacuum without putting some effort into common solutions that others were going to get behind. Or at least not intentionally doing things differently in a way that couldn’t be easily integrated with other systems later on.
And then Linux grew up and became all sexy. Those involved started changing, not necessarily in bad ways, but just different. Suddenly not everyone enjoyed using terminals, typing commands, or having various daemons and services around. They wanted something for “mainstream” users that had all the fluff and shine of other Operating Systems. And so various projects spawned up over the last decade, seemingly out of a necessity, following the “Bazaar is always best” philosophy (which typically also comes with a hefty dose of libertarian-like laissez-faire thinking). New protocols, new interfaces, whole new approaches just kind of happened on us without any real co-ordinated thinking at a much higher level. So now we have many difference folks pulling in different directions. And each year one “Linux” becomes more different than the next “Linux”. Some want compatibility and standards based development (even if it’s lousy at times). Others want “OMG, not some lame standard, pah! we’re the best! just do it!” and for Linux to do its own thing entirely. Neither approach is entirely correct, nor entirely wrong. But we’re not learning from UNIX either.
Anyway. Now would be a very great time for us to take a deep breath and ask ourselves if we want to have the next ten years be like the last. Do we want to continue along a path that is going to increasingly see two, three, or more “solutions” to common Linux issues, with vendors getting behind one or another, and folks criticizing projects for NIH mentality? Or do we want to have a moment of Zen where we realize we’d be better off playing nicely together on a more comprehensive approach to world domination? I personally would like to see industry standards bodies like the Linux Foundation drive a few years of stabilization and standardization wherein we get behind common Linux ways to do things so we don’t turn into the next incarnation of UNIX.
I speak only for myself, etc.
Jon.
I think at least some of this is explained by the necessity of doing things one way N years ago, and the lack of the same today. For instance when I first started to use Linux I _needed_ my code to run on Solaris (most of the servers were Sun boxes), and there were a significant number of *BSD servers too.
But for a long time now the only real question has been “What kind of Linux box is it?” — and even more recently, with VPSs, even that’s not a question as it’s always the arch. (of i686 or x86_64) you want running the OS you want.
So, to put it another way, there was a huge incentive to “be compatible” and now there is almost no incentive to do so.
Great thought provoking post Jon,
I have been a GNU Linux user for some time but still tend to play in the higher level areas. Currently I run Mandriva and maybe soon Magia.
You speak of Fragmentation and the two major issues I have seen in that arena are Beryl vs. Compiz and now Xorg vs. Wayland. Clearly there are other possibly more well known or of bigger issue topics out there but those are the ones that I think of.
Sure Android has its own self made fragmentation issues but for me that is a different topic. Also we may have GNU Linux fragmentation between the Android(s) vs. Meego vs. Ubuntu Unity vs. Chrome OS vs. Whatever Embedded Linux. Here I will stake my live free or die hopes on Meego.
More to what seems to be your point is that the LSB has been said to be a failure but I do not know all of the reasons why. With or with out it I think it has been a good few years for Desktop Linux in general. If only for some name recondition and more common support from hardware manufactures. More good stuff along those lines is that there are a growing number of Linux only hardware vendors. Which because as it is separate from elusive branding and is open source can be great for all OSes. Though Microsoft would never admit such a thing.
Anyway all of that is to restate my Xorg vs. Wayland questions. Wayland looks great for the desktop given time to mature. If given that same time to mature it might be equally great as Xorg is now for the everything that is not a desktop or other graphics heavy environment. Who knows.
So those are my thoughts as scattered as they are thanx for the post.