And you proved my point - to generate interest you've got to have some discussion and controversy.
While discussion is excellent, as it shows an healthy and active community, controversy usually turns off the people because it indicates the existence of underlying unresolvable issues.
Shock marketing attracts attention - although not that necessary to crystallize a vision in a positive atmosphere.
[quote=alexixor]
I really struggle to see the reasoning behind your opinion that âthere is no point to have BeOS/Aros whatever, because it does not have any applicationsâ.[/quote]
If youâre going to quote me, just quote me. Donât put words in my mouth. If you canât find something I actually wrote that reflects what you believe I meant, then you should consider very strongly whether I actually meant it or if itâs just in your head.
I suppose you did not think very hard about this idea, on the contrary itâs usual to release operating systems for which applications already exist. A particularly good illustration is Windows NT 4.0
NT 4.0 primarily ran Win32 programs, the same API exists to this day in Windows 10. But Microsoft went to particular effort to ensure that it was possible for developers to run many applications not only on Windows 95 (which had been intended to ship as Windows 93, several years before NT 4.0) but even on Windows 3.1 through something called Win32s. The Win32s components were royalty free so it was almost as simple, and often preferable, to program a 32-bit program using Win32s than to write a suitably powerful 16-bit program that was directly compatible with Windows 3.x
It so happens that in the Win32 time frame the graphical web browser was just becoming âa thingâ. These were large, complex applications primarily developed on 32-bit platforms. Bringing them to 16-bit Windows wasnât entirely unthinkable but would certainly have meant a large amount of extra effort for minimal reward given how under-powered the Intel 286 processors were. So they targeted Win32s, and for the first few years most of their users will have used Windows 3.1. But because the Win32 APIs were used it meant that on day zero the browsers ran on Microsoftâs brand new Windows NT 4.0, and they ran much better there than in a cramped corner of Windows 3.1
Actually as weâve gone over previously âdesigningâ wasnât part of the brief for Haiku. It is in most respects simply a clone of BeOS, there are lots of places where Haikuâs developers have no explanation (or at least no coherent explanation) for the design, theyâre just copying how Be Inc. did it.
Haiku is a retro-computing OS. It exists because nostalgia is a thing. If you squint you could pretend that for the first few years it was seriously endeavouring to be a natural successor to the extant BeOS R5. But today itâs like any other retro-computing OS. So far as I can see it has far more in common with a non-free retro-computing OS like say, AmigaOS 4 than with free modern operating systems like Debian.
and some comments later:
[quote]
I suppose you did not think very hard about this idea, on the contrary itâs usual to release operating systems for which applications already exist. A particularly good illustration is Windows NT 4.0[/quote]
Erm⌠Be wanted to create a new OS, remember?
OS = Kernel + libraries, so how would it be possible to have a new OS using the NT kernel and/or the Win32 libraries?
Also, it is funny that you seem to think that NT 4.0 was new⌠NT3.5.1 was released on May 30, 1995.
Microsoft did very well in making NT4.0 run older windows software, something Apple almost failed to do with MacOSX 10.0 vis a vis earlier Mac software.
It would be stupid for a vendor to ship an OS that is not backwards compatible with said vendorâs older S/W, no?
But what does this have to do with Be?
Be wanted to create an operating system from scratch. Why would they then add say Win32 compatibility?
How would that attract developers?
By your logic, Red Hat/Suse/etc should ship their distro with a supported wine and advertise that fact as a selling point.
We all know what happened of these efforts, donât we?
If you cannot search, or donât know, take a look at:
I believe you know exactly the reason why reimplementing someone elseâs API is a completely an utterly stupid idea, no?
[quote]
So far as I can see it has far more in common with a non-free retro-computing OS like say, AmigaOS 4 than with free modern operating systems like Debian.[/quote]
Who said that Haiku is modern?
Also, what does âfreeâ, I assume you mean libre here, have to do with what Haiku resembles?
Also, how is Debian âmodernâ?
If by modern you mean H/W drivers, then yes, debian is âmodernâ.
If you mean the userland, including the condescending piece of crap that is Gnome, the hopeless mess that KDE is, or their retarded dysfunctional cousins XFCE/Unity etc, then I suggest you read a little on what a proper designed UI is.
And then try to see how many operating systems had their UI enhanced by a University doing exactly that, UI research.
If by modern you mean the Linux kernel, then I am afraid you are, again, wrong.
Fast, yes.
Flexible, yes. I can boot my watch with it AND run facebook using the same kernel, how cool is that?
Modern, I am afraid, no.
I mean, the volume manager is something that came back to the future from the 80âs, to take just one exampleâŚ
In my book, SmartOS is an example of a âmodernâ OS.
Actually SmartOS is a fine example of a system that is so far beyond of what Linux, and the surrounding distro zoo, can do at the moment, it is not funny any more.
But it is not meant for a desktop.
There is exactly zero desktop software that is Linux specific and anyone sane would use over an alternative.
Haiku should, and does, support QT, if it has to support a cross platform API, because QT was designed to be cross-platform in the first place, but this is neither here nor there.
So from the above, and given the fact that Haiku has billions of dollars less funding compared to âmodernâ Linux, I think the outcome is fairly good and dare I say, more consistent and less buggy compared to the other wannabe âDesktopâ OSen (remember the billions of $$ and man-hours that go into Linux development):
Also, implementing a closed source kernel system call table or API, is futile, because of obvious reasons.
I mean, I love ReactOs, I think it is a very useful project, but if you install it and see the software catalogue, save for firefox and thunderbird, it actually has less software available than HaikuâŚ
So what exactly is your problem with Haiku?
I suppose this is some small progress.
Be âwantedâ Appleâs board to beg JLG to come back, on their hands and knees. Everything else was just part of the show to support that dream.
Youâre the one moving the goal posts here. Your claim was that âThere is no OS that had applications before releaseâ. The problem youâre running into here isnât about subtleties of phrasing, itâs simply that your idea of how this works is completely wrong.
Appleâs transition followed a similar plan, encourage developers to move to an API family called Carbon that can be supported on the old OS but runs better on the new one. Even some of Appleâs own apps were still using Carbon a decade later. However Apple has a very different relationship to their ISVs than Microsoft. Win16 applications (from the 1980s) were still supported on Windows long after Apple had deprecated Carbon.
Youâd have to ask Be Inc. about that, as it was their stated intention to take this approach, clean slate every few years with a new OS and green field development. Crazy plan, but Iâm not the one lauding their approach.
So, âthese effortsâ means a tiny distro few have heard of, signing a deal with Microsoft and then sinking beneath the waves, why did you invoke Red Hat ? In fact Red Hat do very well off heterogeneous systems and if you want to run Windows on RHEL, Red Hat provide and fully support virtio drivers to do that.
Youâre writing this in defence of Haiku, a re-implementation of someone elseâs API?
I understand that youâre not necessarily going to read what I wrote, but you would do well to at least read what you wrote. You specifically cited Haikuâs âopen source techniqueâ, which if it means anything means Free Software.
Plenty of people do UX research. Itâs nice that Haiku got some done for free by outsiders in Auckland, but you rather over-estimate how much thatâs worth. Keep in mind that today, in 2015, the authorised public APIs for Haiku still arenât font sensitive.
You mean the logical volume manager? Thereâs not a tremendous pace of development in that space, but it takes quite some squinting to claim that itâs from the 1980s. If you meant some other component for which you donât remember the actual name, maybe try to find that out before making sweeping comments about it?
[quote]In my book, SmartOS is an example of a âmodernâ OS.
Actually SmartOS is a fine example of a system that is so far beyond of what Linux, and the surrounding distro zoo, can do at the moment, it is not funny any more.[/quote]
Yes, weâve established that you believe this in a previous post. The key feature of SmartOS, of course, is that it has ported the entirety of Linux KVM. Itâs weird that porting part of Linux suddenly makes a system âso far beyond of what Linux can doâ in your view.
Your big idea for what to do with things like SmartOS seems to be⌠run Docker, another Linux native technology. SmartOS can run Docker almost as fast as a native Linux, so long as you have a pure compute workload. If your workload does actual I/O and thus has some kind of consequences then SmartOS is slower, but now who is counting?
This is essentially a No True Scotsman argument since it allows you to insist that the Linux-specific software wouldnât be used over some hypothetical alternative by âanyone saneâ. I suggest instead writing âI really, really hate Linuxâ and, if you donât like the resulting lack of rhetorical force, just deleting the entire paragraph.
Just to remind you again, BeOS was a closed source OS, and the purpose of the Haiku project is to implement those closed source APIs, which youâve just described as futile.
[quote]
Youâre the one moving the goal posts here. Your claim was that âThere is no OS that had applications before releaseâ. The problem youâre running into here isnât about subtleties of phrasing, itâs simply that your idea of how this works is completely wrong.[/quote]
NT was a Microsoft product, so it makes sense for them to implement their API on it.
The same goes for BeOS. BeOS 5 could run software written for BeOS 4 beautifully.
Your mind has weird ideas it seems, because you think that the correct way to go was to use the NT kernel (how? steal it maybe?).
They wanted to create anew OS.
This is probably beyond your ability to comprehend, but it is what they actually wanted to do.
[quote]
So, âthese effortsâ means a tiny distro few have heard of, signing a deal with Microsoft and then sinking beneath the waves, why did you invoke Red Hat ? In fact Red Hat do very well off heterogeneous systems and if you want to run Windows on RHEL, Red Hat provide and fully support virtio drivers to do that.[/quote]
Xandros was a bad idea, so it died off.
Running windows on RHEL in a VM, is not the same thing as implementing the Win32 API.
Please be careful next time and do your homework first.
[quote]
Youâre writing this in defence of Haiku, a re-implementation of someone elseâs API?[/quote]
Haiku is reimplementing a frozen API.
This makes it possible to actually do it, even if it takes 20 years.
Implementng Win32, Quartz etc, while these are alive and evolving is an exercise in futility.
Again look at ReactOS.
[quote]
I understand that youâre not necessarily going to read what I wrote, but you would do well to at least read what you wrote. You specifically cited Haikuâs âopen source techniqueâ, which if it means anything means Free Software.[/quote]
You did a non sequitur comparison of Haiku to AmigaOS, and I (tried) to make sense of it.
Somehow it is not âmodernâ compared to âmodernâ debian, but of course you failed to explain what makes debian modern in the first place.
[quote]
Plenty of people do UX research. Itâs nice that Haiku got some done for free by outsiders in Auckland, but you rather over-estimate how much thatâs worth. Keep in mind that today, in 2015, the authorised public APIs for Haiku still arenât font sensitive[/quote]
We agree then, even in tangent, that it is a good thing.
Gnome 2 had some UI testing too, BTW.
Not as extensive as the Auckland effort, but good nonetheless.
The font sesnsitivity âissueâ is not a fundamental flaw in the way Haiku is put together. It is a lack of time and manpower.
[quote]
You mean the logical volume manager? Thereâs not a tremendous pace of development in that space, but it takes quite some squinting to claim that itâs from the 1980s. If you meant some other component for which you donât remember the actual name, maybe try to find that out before making sweeping comments about it?[/quote]
The Linux distros are competing with UNIX alternatives that have far more advanced capabilities, including âmodernâ ways of managing storage.
But letâs donât be âgenericâ about it and be specific.
How about implementing a proper epoll, say by looking the way that kqueue works, and put epoll out of itâs misery?
Or how about having a filesystem that is actually from this century?
Or maybe proper network virtualization?
Some real namespacing maybe, and not the farce that the Linux hype machine calls âcontainersâ?
Also, please do not give me examples using something else except RHEL and/or Suse, because only supported releases count. You know, the variety of Linux that people are willing to pay real money for.
Also Ubuntu fails by default, thanks.
[quote]
Yes, weâve established that you believe this in a previous post. The key feature of SmartOS, of course, is that it has ported the entirety of Linux KVM. Itâs weird that porting part of Linux suddenly makes a system âso far beyond of what Linux can doâ in your view.
Your big idea for what to do with things like SmartOS seems to be⌠run Docker, another Linux native technology. SmartOS can run Docker almost as fast as a native Linux, so long as you have a pure compute workload. If your workload does actual I/O and thus has some kind of consequences then SmartOS is slower, but now who is counting?[/quote]
SmartOS is meant to be used to run software that also runs on Linux.
So it makes sense they ported kvm and that they implemented the Linux system call table.
To run Linux binaries, including Docker of course.
But SmartOS can run Linux binaries at least as fast as Linux itself, plus have all the other good stuff, say ZFS/Crossbow/Dtrace etc
This is their selling point to the Linux admins.
The people who actually know what they are doing would not use Linux in the first place, but no one is perfect and you learn something new each day. 
Linux has contributed zero new ideas and software since itâs inception,apart from proper package management and the popularization of the open source way to create software.
These have nothing to do with Linux the kernel, or the distributions.
As a matter of fact Linux is not a very capable system in itself at all if you look closely.
In fact Linux has tried, and failed for the most part, to re-create (ape I would say), most of the technologies modern OSes have.
Examples are (but are not limited to):
Dtrace<â>systemtap
BtrFS<â>ZFS
OpenVZ/cgroups/etc<â>jails/zones
Liveupgarde<â> ksplice etc
Every single one of these has been either a marginal success, ksplice, or a bad joke, systemtap/BtrFS etc.
I will not even go to network virtualization, because it will be just too easy.
I mean try to give a docker âcontainerâ a real interface in RHEL.
Another example of an OS that is created to run another OS is XinuOS:
The purpose if life of this OS is to run an Os and software from a dead company too, SCO.
There is a huge market for this, since the days of the mainframe.
The benefit of course it that you cna run your old crappy OS/Software, using facilities of today, ZFS/Dtrace etc.
Of course Haiku does not want to run anybody elseâs software, apart from some old BeOS programs, and this is exactly what is does.
[quote]
This is essentially a No True Scotsman argument since it allows you to insist that the Linux-specific software wouldnât be used over some hypothetical alternative by âanyone saneâ. I suggest instead writing âI really, really hate Linuxâ and, if you donât like the resulting lack of rhetorical force, just deleting the entire paragraph[/quote]
Name one piece of desktop software that originated in Linux and was ported over to another platform because either there was no better alternative or that people all the alternatives where simply inferior:
Examples of desktop software are:
A word processor.
A text editor.
An image editor.
I can certainly think of at least one desktop program that was created on BeOS and was ported to at least three platforms due to popular demand.
I will give you a cookie if you find it, promise.
[quote]
Just to remind you again, BeOS was a closed source OS, and the purpose of the Haiku project is to implement those closed source APIs, which youâve just described as futile.[/quote]
See above, the part where I explain in simple English that the BeOS API is frozen, by virtue of the fact that Be is dead dead
Iâm sorry, but I can not agree with that assessment. They clearly had time to write a browser, email program, media player, etc etc etc. But they also gave you the code with most of those programs, and had no problem letting and helping people write replacement programs for the desk bar and tracker etc. Yes, Iâm sure that happens a lot, where people say I wanted to do this, that way, etc. But its a little cynical to say everyone works that way
But then again, when Dano was leaked, it did seem like they were moving on from super simple to something a little deeper, little fancier, little more CPU heavy
Everything is this world evolves, grows, changes, etc. There is a point in the evolution of everything, where it gets too big, too old, too complicated etc. In nature/life, those things die, in TV shows, they call it jumping the shark, etc. In OSâs, they call it Windows, lol
Ive been using Linix MINT lately, I really like it, its great in most ways, and with Linux, I really do understand the need for a package manager, with thousands of free programs out there, you really need one. Unlike with Windows, where you do not need one, because most programs cost money, and most people only ever use 2 or 3 different programs
These days, you check your email, log into facebook or myspace, etc. watch some porn, etc. And that about it. You really dont need much more. Unless your at work, and then pof course, you use office etc
BeOS, didnt have enough software out there to need a package manager, Bebits was great for finding anything and everything you could want. And I cant see Haiku having much to choose from either
Also, and this may be completely wrong, but if Haiku does make it, I would really like to see the people behind it, become rich, make money, get a little bill gates treatment
I know its open, and that is great, and the way it should be, etc etc etc
But I also think people should get paid well for hard work. I guess Iâm for a open, but closed at the core, kind of OS / software. Where the people can get what they deserve
I hate paying $100 or more for Windows, and love Linux MINT for free etc
But it would be nice to see a real OS, that worked and was mature, just sell cheap, $25 or something, then they could also make office applications that they could sell, as long as they shared and stayed translucent etc. That would be great for everyone
The Haiku programmers could quit their day jobs, and devote more time to the core OS
Anyways, best of luck to all of you, I really hope Haiku makes it, but I just dont see it happening any longer⌠if you can make BEOS live again, you will be gods in my eyes. But at this point, I think it will be just another version, or something that looks and works just like Linux, a great free OS, that most people still just dont want
FREE, is not what people are looking for, simple, secure, easy, fast, not complicated, not over packed / bloated, not an hour install, not trying to guess at what you want next
Actually, thats what I miss most about BE
When you put a CD in, a CD icon appeared on your desktop
You could choose to click on it, or ignore it
No program popped up to ask you what to do
No stupid Windows, searching the CD, and interrupting what you were doing
No guessing at what you wanted to do next
It simply let you know, a CD was now in the drive
Its that simple idea, no one else has ever seemed to latch on to
At least not without making you do extra work, to turn those features off
[quote=alexixor][quote]
Your mind has weird ideas it seems, because you think that the correct way to go was to use the NT kernel (how? steal it maybe?).
[/quote]
Five years ago I wrote in a completely separate context that NT would have been a good choice âif it was availableâ. The rest is a product of your fevered imagination.
[quote]
Xandros was a bad idea, so it died off.
Running windows on RHEL in a VM, is not the same thing as implementing the Win32 API.
Please be careful next time and do your homework first.[/quote]
To reiterate, the point of the operating system is to run applications. Not to make a big show of how many wheels youâve re-invented, just to run applications.
Itâs very simple. Haiku is retro-computing, just like AmigaOS. The Amigans keep up the same charade, they even have people who keep arguing Amiga, remaining stuck in the early 1990s (at best), is truly âinnovativeâ while the systems that moved on are instead left behind. I guess it makes the people who argue this way feel better about themselves? Hard to tell, they sure are angry.
In one way theyâre doing better than Haiku, they have a âsugar daddyâ pouring hundreds of thousands of dollars worth of capital into the ecosystem to try to keep things afloat.
Every project can say theyâd do more if they had the money and time. The fact remains that 3rd parties on Haiku are still, in 2015, offered this awful Petzold-style co-ordinate API instead of a proper UI layout API.
It doesnât really matter which flavour of API you choose here, even completion ports are fine.
To be clear, this is specifically the âhype machineâ you were on earlier with your claims about Docker?
Even the SmartOS developers themselves do not believe this. A fan might be confused though because SmartOS articles tend to quietly shift from talking about SmartOS being able to run Linux binaries to talking about comparisons between running a ported app directly on SmartOS (ie now itâs a native Solaris app) and running a Linux app under KVM. No surprises which of those is fasterâŚ
These âpeople who actually know what they are doingâ decided that the only way forward was to port a significant Linux subsystem. Not re-create, not come up with their own superior alternative, but just port the whole thing. Maybe they know something you donât?
In a sense there are no new ideas. If you squint hard enough you can always find somebody else, with a sort-of similar idea from a little earlier. Yet somehow here we are in 2015, with smartphones and celebrity vloggers, not Pong and BASIC listings magazines. Linux isnât the first kernel to do KMS, and itâs not the first POSIX implementation to have 1:1 kernel threads, and itâs not the first PC software to offer full disk encryption. You can always find a way to prove somebody else did it first, no matter what âitâ is, but still the actual results are hard to argue with.
[quote]As a matter of fact Linux is not a very capable system in itself at all if you look closely.
In fact Linux has tried, and failed for the most part, to re-create (ape I would say), most of the technologies modern OSes have.[/quote]
âŚand then you listed basically a bunch of Solaris features. I get it, youâve decided this week you really like SmartOS. Thatâs nice. Really. But this isnât âmost of the technologies modern OSes haveâ, itâs the set of big name features for Solaris and unsurprisingly Solaris fans think theyâre great.
[quote]
I can certainly think of at least one desktop program that was created on BeOS and was ported to at least three platforms due to popular demand.[/quote]
I recommend thinking very carefully about what âpopular demandâ means in this context. Amigans are currently celebrating the enormous âpopular demandâ that has lead to several hundred people downloading a free program.
[quote]
Five years ago I wrote in a completely separate context that NT would have been a good choice âif it was availableâ.[/quote]
We agree then.
[quote]
To reiterate, the point of the operating system is to run applications. Not to make a big show of how many wheels youâve re-invented, just to run applications.[/quote]
Agreed.
Haikuâs purpose in life is to run BeOS applications and if possible to run new applications.
BeOS applications run fine in Haiku, so mission accomplished.
If the Haiku developers wanted Haiku to run Amiga applications, they would do that, no?
RHEL users want to run Windows in a VM.
I donât see where the confusion comes form?
[quote]
Every project can say theyâd do more if they had the money and time. The fact remains that 3rd parties on Haiku are still, in 2015, offered this awful Petzold-style co-ordinate API instead of a proper UI layout API.[/quote]
Yup, but you you love to spin this so that it looks like it is impossible to do this, or very hard, because the way Haiku is put together is fundamentally flawed.
[quote]
To be clear, this is specifically the âhype machineâ you were on earlier with your claims about Docker?[/quote]
Heh, if you donât understand why Docker is used interchangeably with the word containers, then I will not make it easy for you.
[quote]
Even the SmartOS developers themselves do not believe this. A fan might be confused though because SmartOS articles tend to quietly shift from talking about SmartOS being able to run Linux binaries to talking about comparisons between running a ported app directly on SmartOS (ie now itâs a native Solaris app) and running a Linux app under KVM. No surprises which of those is fasterâŚ[/quote]
You have things a bit confused here, but it is all right, I choose to help you out.
As you keep saying, and I agree of course, the point of an OS is to run applications.
SmartOS runs Linux binaries directly, just like you can do with say wine.
But also you can âbootâ a complete centos system, or ubuntu if you prefer, and use it as if it was a VM.
So there is no actual porting of applications. I mean, they do not port apache or yum, they just run it directly on the metal.
They run unmodified on the SmartOS/Illumos kernel, which happens to be quite fast indeed.
But SmartOS is a very advanced hypervisor, so it is logical to try to make it run as many OSes and applications as possible.
I would be baffled if Joyent started writing SmartOS exclusive softwareâŚ
[quote]
These âpeople who actually know what they are doingâ decided that the only way forward was to port a significant Linux subsystem. Not re-create, not come up with their own superior alternative, but just port the whole thing. Maybe they know something you donât?[/quote]
The reality is that there is a huge body of âLinuxâ software, which I think is not precise, because what is actually out there are very popular distributions that have packaged open source software.
This is what people want, good package management. And this is what the distros deliver.
So now that you can âbootâ ubuntu directly on SmartOS, or Illumos for that matter, why not port docker too?
And now, what do you need Linux for? Docker runs on Windows, MacOSX, Linux, etc.
So in a couple of years when every linux application will be a Docker âcontainerâ, why not choose the best platform that speaks docker?
Do you want to bet that Linux, or better RHEL/Ubuntu etc, are the best platforms to run Docker?
Docker might have started life in Linux, but the way things are going, docker will probably kill Linux off.
[quote]
In a sense there are no new ideas. If you squint hard enough you can always find somebody else, with a sort-of similar idea from a little earlier. Yet somehow here we are in 2015, with smartphones and celebrity vloggers, not Pong and BASIC listings magazines. Linux isnât the first kernel to do KMS, and itâs not the first POSIX implementation to have 1:1 kernel threads, and itâs not the first PC software to offer full disk encryption. You can always find a way to prove somebody else did it first, no matter what âitâ is, but still the actual results are hard to argue with.[/quote]
Linux specifically has contributed the following in computing:
There has been zero new OS development in Linux.
It was created as a way to replicate, on the PC, functionality that was available in UNIX.
SUN was the last company that did real R&D on fundamental OS technologies, at least in *NIX.
The fruits of this efforts are the known ZFS etc.
Linux has tried, and failed due to the complete chaotic way it is developed, to replicate the newer technologies.
[quote]
âŚand then you listed basically a bunch of Solaris features. I get it, youâve decided this week you really like SmartOS. Thatâs nice. Really. But this isnât âmost of the technologies modern OSes haveâ, itâs the set of big name features for Solaris and unsurprisingly Solaris fans think theyâre great.[/quote]
I am correct and you know it.
There is no fanboyism in noticing that Linux is trying to replicate every single one of these technologies.
Also, it is not an accident that Red Hat will not support âfull OSâ containers of any kind, not even LXC. The reality is that they would love to be able to do it, but they cannot because of the incomplete namespacing support in Linux, among other deficencies.
The same applies for the btrfs situation, systemtap, openvswitch etc.
Heh, popular demand was wrong wording.
No. SmartOS has a branded âzoneâ that can run some Linux binaries. Itâs actually left over from when this was how Solaris was going to crush Linux years ago. Remember that? Oh right, it never went anywhere.
The SmartOS version of KVM provides only a subset of the Linux functionality that Joyent felt addressed their philosophy of how to sell cloud services. If you donât happen to agree precisely with that philosophy, too bad.
Itâs Solaris, just minus the branding. It still runs all the same Solaris software now under the IllumOS name. And indeed people still port software to it. Joyentâs âsecret sauceâ runs on SmartOS and nothing else.
I actually ran into a developer who didnât grok what role Linux plays in their âWindows Dockerâ system the other week, it was quite fun to get them to slow down and take things step by step until they have the heart-breaking moment where they realise, hey, this isnât inside a container, why is this Linux?
When âDocker runs on Windowsâ or âDocker runs on MacOSXâ whatâs actually happening is that you run a piece of software called âboot2dockerâ. This is a virtual machine, running Linux. Itâs not a very heavyweight Linux system, just barely enough to run Docker, and so you lose out on a lot, but if youâre just experimenting with Docker and donât yet care about more than the basics itâs pretty impressive.
But as to âwhat do you need Linux for?â well the answer turns out to be âeverythingâ in this context.
Unless youâve got a strange definition of best thatâs easy. Sure, itâs a pretty safe bet. Itâs even the bet made by Joyent - maybe you werenât paying enough attention. Theyâre actually reliant on Linux ABI compatibility which they like to pretend is this weird new thing and say things like âTorvalds found religionâ which is odd because in fact such compatibility is there from the outset. But it didnât suit Solaris people to believe in it ten years ago, so now that it does suit them theyâve got to pretend thatâs not because theyâre slow on the uptake, got to be somebody else. In ten years time theyâll be telling you the same about whatever new long understood thing theyâve just accepted as true.
When the fan sees a rainbow they notice only one shade and declare âAh, my favourite colour, how did you know?â. Itâs not just your Solaris features, you can name almost any oddball feature and somebody was doing it on Linux. Some are now really important, most are largely irrelevant and forgotten. Trying to use just the âworkingâ bits of a damaged DRAM stick without losing data? Linux has that. IA64 (âItaniumâ) support? Linux has that. N:M threading? You could try it on Linux. âUniversalâ drivers for network devices invented by Microsoft? Terrible idea but Linux can run them. Direct access to 520 byte blocks on spinning rust? Linux can do that. Flash RAM as storage without a fake âblock deviceâ or SCSI / SATA interface? Linux can do it. In-kernel HTTP server? Linux. Running as a micro-kernel âserverâ? Linux tried that too.
Not dead. I just tried the iso a few months ago. And think about this, alternative operating systems are good to balance against the other commonly used systems (Windows,Mac OS that have become bloated, full of spyware, advertising and data collection. On top of that, their user interface design is horrible. The last Windows that was practical to use and had a good design was XP (with Windows Classic theme). I never bothered to use Luna or Homestead etc because it was too garish. At least the greys helped keep you focused on work when you have to stare at the screen for long hours. Haiku and BeOS uses these type of greyâs (with some blue and yellow in the default theme) that wonât tire your eyes. Plus, blue and yellow go well together.
This thread is 8 years old. It was made during the âdark agesâ between alpha4 and beta1. I think thereâs not much reason to resurrect it.