![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
lprovenSo, if you follow these things, Oracle has launched "Unbreakable Linux". What
it has done is download all the sources that RH makes publically available for
RHEL, gone through them & removed all the RH trademarks, and recompiled
it, to producea binary-compatible distro with the names and serial numbers
filed off. This is nothing new; several projects have already done this, of which
CentOS is the best known.
I must admit that RH's movements a couple of years back rather mystified me -
killing off the free product, offering a sort of rolling testbed instead in the form
of Fedora, and demanding $LOTS for all versions of RHEL, but hey, it seems to
be working for them. I half-suspect that Oracle's mucking about with
Oracle Linux is more of a warning shot across RH's bows than anything
else.
But saying that, and even allowing for the fact that ultimately the whole
model of corporates making money from FOSS &c is all new and unproven, I
don't have a lot of faith in the ways that either RH or SuSE are doing it.
I think that I grok why and how and what Canonical and Ubuntu are doing,
but that's one rich man's fiat, really. It's different.
But one thing, I suspect, is relatively certain. There's a lot of
consolidation yet to happen in the FOSS world. Some of that will come from
M&A activity, some will be companies simply going out of business.
My guesses?
My personal money is on, medium term, Linux in specific and FOSS in
general doing a huge amount of damage to the commercial software world.
Every commentator I've read who says that it won't or that it's not
important fundamentally Does Not Get It AFAICS.
I don't think FOSS will kill closed-source S/W, ever, but I think it will
eventually force it into small niche markets and into private development
inside companies.
I reckon that in the medium to long term, GPL stuff will beat out all
other FOSS except for a few individual major products, e.g., X.org.
I suspect that until something better than Unix comes along, Linux will in
the end beat out all other forms of Unix. However, I do feel that Unix
itself is a bit long in the tooth.
I reckon that its replacement will be gradual and almost imperceptible at
first. My vague intuition is that it will be something that's close to
Unix at first, but gradually diverges. Look at Mac OS X. It's Mach +
FreeBSD + OpenStep + PDF + OpenGL + lots of other big chunks of FOSS. The
result is something that runs rings around any and all other desktop OSs.
But it's a SOTA WIMP GUI desktop, it's not a killer server OS or
thin/small client OS or a NOS. I really Do Not Want a palmtop running OS
X, any more than I want one running XP or Linux!
I think there are lessons - some specific, some general - to be learned
from BeOS, from AmigaOS, from the NeXTstep family, from Plan 9, from Minix
and Amoeba, and so on and so on.
To make some more specific predictions:
Kernels are going to change. Things like GNU kFreeBSD and GNU kNetBSD and
Nexenta and so on will gradually split the GNU userland away from
dependance on any particular kernel. The FOSS Unix userland will evolve in
some interesting new ways. I think X will blossom then fade away, replaced
by nearly-compatible newer efforts with clever advantages. Init is going
to change, as are desktops and GUI models in general. The sharp edges will
get worn off the various bits of FOSS Unix, so that you can mix & match -
with non-trivial effort - different kernels (Linux, xBSD, Solaris, various
weird microkernels or message-passing modular kernels) with your preferred
init replacement and you preferred GUI, but on top of this, you get the
whole layer of apps and services running on pretty much anything.
I can't believe that huge monolithic kernels are going to keep growing and
improving. I don't think the unaided human mind can improve such huge
monsters much. I'd also bet that multi-multi-multi-core PCs, not SMP in
design, will gradually force kernels to evolve into... well, herds of
Unix-replacing daemons. ;-)
Unix will spread and diversify into a collection of loosely-related,
largely-compatible OSs, which share enough characteristics to share the
same userland and the same apps, but which are targetted at different
functions, from smartphones to workstations to servers to huge cluster
supercomputers.
And that this adaptive radiative pressure will favour modular code that
comes in chunks that small teams and individuals can comprehend and
master.
And that up against all this, private companies with monolithic OSs based
on closed code will just not be able to compete.
Today's model of big general-purpose machines running general-purpose OSs
and mix'n'match big complex apps will be increasingly pressurised by
clever little convergence devices coming up from below, doing more and
more and doing it in simple, integrated fashion. The GP desktop computer
is destined to go the way of the high-end Unix workstation: not extinct,
but in a niche.
However, the techniques and technologies that make it tick will be
absorbed, piecemeal, by little mobile battery-powered radio-communicating
gadgets. The lines will blur between games consoles, both home and
handheld, music players, phones both landline and pocket, cameras, PDAs,
laptops, video recorders, TV sets and so on. Forgotten or unsuccessful
tech like chording keyboards, true handwriting and speech and gesture and
facial expression recognition, and things like that, might well reappear,
soon, as these devices will see more innovation than has happened in 2
decades on the desktop.
At first, in these gizmos, FOSS will not be a big presence, but it will
eventually catch up. On the desktop, it will rule.
Sorry.
Wandered off into ranting prophet mode there.
Some more specific guesses:
OS development requires significant investments of resources. It will stay
the ballpark of big, rich companies.
Companies which are diverse, which want fingers in all markets, will get
out of OS development. The giants will buy the medium-size independents,
the little ones will be gobbled up, then ultimately, the lure of selling
on rivals' platforms will mean the OS divisions will get spun out.
Surprising and unpredictable things will happen to the Big Iron vendors
who make hardware and sell their own OSs. Ultimately these 2 things will
prove incompatible with one another if the company wants to play in the
open market. So, say, the tightly-integrated Apple model might continue.
There will be big mergers and subsequent splits among the giants: Sun, HP,
IBM, Oracle. Companies who make hardware that runs other companies' OSs
will end up getting out of the OS market themselves.
There's strength in owning your own exclusive combination of the 2, but
it's precarious, but if the pairing is non-exclusive, it's a weakness.
To keep open systems open, I feel that the only way is for there to be
deep separation between OS developers, app developers and hardware
developers.
Hardware wise, soon there will be 2 main platforms: ARM and x86. One will
continue to get more frugal, the other will continue to absorb the rest of
the system and gain performance through integration of "peripheral"
devices.
Everything else will get marginalized.
Virtual environments like Java and .NET/Mono will blur the distinction,
anyway.
it has done is download all the sources that RH makes publically available for
RHEL, gone through them & removed all the RH trademarks, and recompiled
it, to producea binary-compatible distro with the names and serial numbers
filed off. This is nothing new; several projects have already done this, of which
CentOS is the best known.
I must admit that RH's movements a couple of years back rather mystified me -
killing off the free product, offering a sort of rolling testbed instead in the form
of Fedora, and demanding $LOTS for all versions of RHEL, but hey, it seems to
be working for them. I half-suspect that Oracle's mucking about with
Oracle Linux is more of a warning shot across RH's bows than anything
else.
But saying that, and even allowing for the fact that ultimately the whole
model of corporates making money from FOSS &c is all new and unproven, I
don't have a lot of faith in the ways that either RH or SuSE are doing it.
I think that I grok why and how and what Canonical and Ubuntu are doing,
but that's one rich man's fiat, really. It's different.
But one thing, I suspect, is relatively certain. There's a lot of
consolidation yet to happen in the FOSS world. Some of that will come from
M&A activity, some will be companies simply going out of business.
My guesses?
My personal money is on, medium term, Linux in specific and FOSS in
general doing a huge amount of damage to the commercial software world.
Every commentator I've read who says that it won't or that it's not
important fundamentally Does Not Get It AFAICS.
I don't think FOSS will kill closed-source S/W, ever, but I think it will
eventually force it into small niche markets and into private development
inside companies.
I reckon that in the medium to long term, GPL stuff will beat out all
other FOSS except for a few individual major products, e.g., X.org.
I suspect that until something better than Unix comes along, Linux will in
the end beat out all other forms of Unix. However, I do feel that Unix
itself is a bit long in the tooth.
I reckon that its replacement will be gradual and almost imperceptible at
first. My vague intuition is that it will be something that's close to
Unix at first, but gradually diverges. Look at Mac OS X. It's Mach +
FreeBSD + OpenStep + PDF + OpenGL + lots of other big chunks of FOSS. The
result is something that runs rings around any and all other desktop OSs.
But it's a SOTA WIMP GUI desktop, it's not a killer server OS or
thin/small client OS or a NOS. I really Do Not Want a palmtop running OS
X, any more than I want one running XP or Linux!
I think there are lessons - some specific, some general - to be learned
from BeOS, from AmigaOS, from the NeXTstep family, from Plan 9, from Minix
and Amoeba, and so on and so on.
To make some more specific predictions:
Kernels are going to change. Things like GNU kFreeBSD and GNU kNetBSD and
Nexenta and so on will gradually split the GNU userland away from
dependance on any particular kernel. The FOSS Unix userland will evolve in
some interesting new ways. I think X will blossom then fade away, replaced
by nearly-compatible newer efforts with clever advantages. Init is going
to change, as are desktops and GUI models in general. The sharp edges will
get worn off the various bits of FOSS Unix, so that you can mix & match -
with non-trivial effort - different kernels (Linux, xBSD, Solaris, various
weird microkernels or message-passing modular kernels) with your preferred
init replacement and you preferred GUI, but on top of this, you get the
whole layer of apps and services running on pretty much anything.
I can't believe that huge monolithic kernels are going to keep growing and
improving. I don't think the unaided human mind can improve such huge
monsters much. I'd also bet that multi-multi-multi-core PCs, not SMP in
design, will gradually force kernels to evolve into... well, herds of
Unix-replacing daemons. ;-)
Unix will spread and diversify into a collection of loosely-related,
largely-compatible OSs, which share enough characteristics to share the
same userland and the same apps, but which are targetted at different
functions, from smartphones to workstations to servers to huge cluster
supercomputers.
And that this adaptive radiative pressure will favour modular code that
comes in chunks that small teams and individuals can comprehend and
master.
And that up against all this, private companies with monolithic OSs based
on closed code will just not be able to compete.
Today's model of big general-purpose machines running general-purpose OSs
and mix'n'match big complex apps will be increasingly pressurised by
clever little convergence devices coming up from below, doing more and
more and doing it in simple, integrated fashion. The GP desktop computer
is destined to go the way of the high-end Unix workstation: not extinct,
but in a niche.
However, the techniques and technologies that make it tick will be
absorbed, piecemeal, by little mobile battery-powered radio-communicating
gadgets. The lines will blur between games consoles, both home and
handheld, music players, phones both landline and pocket, cameras, PDAs,
laptops, video recorders, TV sets and so on. Forgotten or unsuccessful
tech like chording keyboards, true handwriting and speech and gesture and
facial expression recognition, and things like that, might well reappear,
soon, as these devices will see more innovation than has happened in 2
decades on the desktop.
At first, in these gizmos, FOSS will not be a big presence, but it will
eventually catch up. On the desktop, it will rule.
Sorry.
Wandered off into ranting prophet mode there.
Some more specific guesses:
OS development requires significant investments of resources. It will stay
the ballpark of big, rich companies.
Companies which are diverse, which want fingers in all markets, will get
out of OS development. The giants will buy the medium-size independents,
the little ones will be gobbled up, then ultimately, the lure of selling
on rivals' platforms will mean the OS divisions will get spun out.
Surprising and unpredictable things will happen to the Big Iron vendors
who make hardware and sell their own OSs. Ultimately these 2 things will
prove incompatible with one another if the company wants to play in the
open market. So, say, the tightly-integrated Apple model might continue.
There will be big mergers and subsequent splits among the giants: Sun, HP,
IBM, Oracle. Companies who make hardware that runs other companies' OSs
will end up getting out of the OS market themselves.
There's strength in owning your own exclusive combination of the 2, but
it's precarious, but if the pairing is non-exclusive, it's a weakness.
To keep open systems open, I feel that the only way is for there to be
deep separation between OS developers, app developers and hardware
developers.
Hardware wise, soon there will be 2 main platforms: ARM and x86. One will
continue to get more frugal, the other will continue to absorb the rest of
the system and gain performance through integration of "peripheral"
devices.
Everything else will get marginalized.
Virtual environments like Java and .NET/Mono will blur the distinction,
anyway.
