[The Godfather]

In 1984 came the Macintosh with the black and white Multi-System-whatever. The case design was square, to match the squarish lines of the finder, and the surfaces were flat so that the hardware wouldn't outstage the software.

With the Color Classic, the OS was still squarish and flat-colored (but colored nonetheless), but the case started to get rounder and rounder. The representative example would be huggable Performa, while some other cases starkly jumped off the on going trend (Pizza box).

Then at one point in time I don't remember, the MacOS UI becomes Platinum. The difference between the OS 7 and OS 8 is a perfect parallel with the "Classic Beige" boxes and the "New Beige" of the then current Apple offerings.

Then came the iMac (and this history is recent enough for you to follow through), with a revolutionary case made of translucent polycarbonate, with a look found nowhere except in some 70's wardrobes. Who on their right mind would imagine that the iMac was a hint to important things to come.

When OS X was revealed, with a brand-spanking new UI, it was received with praise and resentment. "Are we prepared for a Playskool UI?", the sceptics wrote in their blogs. For better or worse, their dolor subsided eventually.

And then came Titanium and Aluminum Powerbooks, and steel G5 and Mac Pros, and OS X was decidedly entering a phase of brushed metal, phasing out the lickable Aqua. Can we say "The Mac Iron Age"?

Focusing on the present, we have iPods with smooth plastic and metals, iTunes that would better match a Sun Solaris workstation, and iPod nanos going back to brushed Aluminum.

What predictions can be drawn on the future of OS X? Or can any conclusion be drawn?

[OwlBoy]

Hmm, this is an interesting look at the whole subject.

It does seem a bit like the hardware does lead the software. At least in the past 6 years or so.

I would say it is safe to assume that the Mac UI will follow the hardware trends and become more sleek and refined.

-Owl

[Buckaroo]

I didn't like the first iBooks. the blue, pink clam shapped ones.

[rickey939]

Originally Posted by Buckaroo

I didn't like the first iBooks. the blue, pink clam shapped ones.

Pink? They were orange weren't they?

[Chuckit]

The answer to the question in the subject is: Yes.

[mitchell_pgh]

A little off topic, but I've always felt that the OS and the Processors have been leepfrogging on another.

We get a nice OS, but our processors are slow... finally get nice processors and the OS is slow.

I'm finally seeing us getting to the point where we have fast processors and a fast OS. I only hope the OS speeds up a bit more.

[goMac]

Originally Posted by mitchell_pgh

We get a nice OS, but our processors are slow... finally get nice processors and the OS is slow.

I'm finally seeing us getting to the point where we have fast processors and a fast OS. I only hope the OS speeds up a bit more.

OS X is actually pretty competitive in terms of speed with Windows.

[Chuckit]

Yeah, I don't think OS X is very slow. Some applications are kind of slow, but there ain't much the OS can do about that.

[goMac]

Originally Posted by Chuckit

Yeah, I don't think OS X is very slow. Some applications are kind of slow, but there ain't much the OS can do about that.

The kernel does have some issues with starting applications, but I'd say that Windows takes a lot of short cuts in this regard that leave it more open to attack. Linux... well... Linux is a kernel that has to be recompiled to add wireless networking support. That should speak for itself.

There is no free lunch in operating systems.

[bradoesch]

Originally Posted by Buckaroo

I didn't like the first iBooks. the blue, pink clam shapped ones.

Apple refused colours and insisted on flavours. The first iBooks were blueberry and tangerine.

[The Godfather]

With MS shooting themselves on the foot with the Vista user agreements, I think it is pretty safe that OSX will continue to gain marketshare, regardless of its sorely loathed shortcomings.

[The Godfather]

Originally Posted by Chuckit

The answer to the question in the subject is: Yes.

Yeah, but can we predict if OS XI will look like iPods, like iTunes, or like Steve blue jeans?

[goMac]

Originally Posted by The Godfather

With MS shooting themselves on the foot with the Vista user agreements, I think it is pretty safe that OSX will continue to gain marketshare, regardless of its sorely loathed shortcomings.

Not to mention Vista's software compatibility sucks. You'll see organizations, that by being forced to rebuy the software they've used for 15 years, consider Macs as replacement machines.

[Kerrigan³]

Has anyone noticed how iTunes looks like Platinum?

[goMac]

Originally Posted by Kerrigan³

Has anyone noticed how iTunes looks like Platinum?

Maybe Apple has finally finished Copland and that's what they'll replace OS X with.

[Goldfinger]

Originally Posted by The Godfather

With MS shooting themselves on the foot with the Vista user agreements

How is that ?

[- - e r i k - -]

One minor fault here: Brushed metal was introduced with QuickTime 5 back in OS 9

[mitchell_pgh]

Originally Posted by Goldfinger

How is that ?

http://it.slashdot.org/article.pl?sid=06/10/12/2240214

This is going to upset some of the geeks out there...

I'm guessing they will change this, but who knows...

[turtle777]

Originally Posted by mitchell_pgh

Slashdot | Vista Licenses Limit OS Transfers, Ban VM Use
.

M$ still thinks it can control the market. They are in for a rude awakening.

-t

[Strupat]

How could an OS possibly be ahead of its hardware?

[Hash]

Simply, you got vision of it, but not the hardware for it, like OS X in 2001 with G3 processors and 32 mb of video ram.

The author has a point, really. Both hardware design and OS design have something in common - in this case, perhaps some graphic design trends. It is hard to say how the Leopard will look like, the main thing is that will perhaps fix GUI inconsistensies and finally say goodbye to Aqua, instead introducing a darker theme. All in all, OS X has been way too bright: you know that design while being esthetically pleasing should not get in the way when working but Aqua was way too fashion-oriented for it. Glad to see it coming with iTunes 7 interface.

[SirCastor]

it may be valuable to note, in this discussion, that hardware design and application is significantly more expensive than the 'theme' of the software/OS. I don't want to say UI, because UI design can be really expensive, but a theme or the look of the OS can be changed with just a little work. Hardware design though takes a lot of money to draw up, prototype, test, redesign, re-prototype and finally produce...

[besson3c]

Originally Posted by goMac

The kernel does have some issues with starting applications, but I'd say that Windows takes a lot of short cuts in this regard that leave it more open to attack. Linux... well... Linux is a kernel that has to be recompiled to add wireless networking support. That should speak for itself.

There is no free lunch in operating systems.

What does wireless networking support have to do with kernel performance? The Linux kernel performs quite well. The fact that it may or may not include wireless networking support simply speaks to more of a strategic/road map type decision than a design decision that would affect overall performance.

[goMac]

Originally Posted by besson3c

What does wireless networking support have to do with kernel performance? The Linux kernel performs quite well. The fact that it may or may not include wireless networking support simply speaks to more of a strategic/road map type decision than a design decision that would affect overall performance.

Yes, Linux performs well, but it's not user friendly. Linux is great for dedicated tasks, but for a normal user, it's not a very good kernel. Can you imagine that?

"Alright honey, I just got the new wireless card installed in our machine. Let's recompile the kernel to get it working!"

Of course in Mac OS X you can just add kernel extensions to extend the kernel without actually having to recompile it.

Hence the "There is no free lunch in operating systems."

[SirCastor]

Originally Posted by goMac

"Alright honey, I just got the new wireless card installed in our machine. Let's recompile the kernel to get it working!"

Many geeks would love a spouse like this...

Actually, I'm kinda glad my GF doesn't know this stuff. At least she'll keep me around for something

[besson3c]

Originally Posted by goMac

Yes, Linux performs well, but it's not user friendly. Linux is great for dedicated tasks, but for a normal user, it's not a very good kernel. Can you imagine that?

"Alright honey, I just got the new wireless card installed in our machine. Let's recompile the kernel to get it working!"

Of course in Mac OS X you can just add kernel extensions to extend the kernel without actually having to recompile it.

Hence the "There is no free lunch in operating systems."

This makes no sense...

Because a certain distro doesn't bundle in a kernel extension for a wireless device, it is a bad kernel design?

Why do you have these knee-jerk reactionary responses to any of my posts involving Linux? We've been over this countless times, I'm tired of making concessions that Linux is not a user-friendly Desktop OS, we all know this now. I've never said otherwise.

Maybe you simply aren't saying exactly what you mean? How about "X distro is no good, or Linux as a whole is no good as a Desktop OS, because kernel extensions for wireless devices are not included"?

Because OS X includes wireless device support really has no bearing on the design and efficiency of the kernel.

[goMac]

Originally Posted by besson3c

This makes no sense...

Because a certain distro doesn't bundle in a kernel extension for a wireless device, it is a bad kernel design?

Yes it does. Because if Linux was designed like OS X, it wouldn't matter what distro you're running. Just install the drivers for your wireless card, and you're done.

Originally Posted by besson3c

Because OS X includes wireless device support really has no bearing on the design and efficiency of the kernel.

An operating system only performs as fast as you can use it. If I have to spend my time maintaining my operating system's kernel, that makes me less productive and my operating system less productive.

Using Linux on a server that is not going to be changing very much is one thing, but using Linux on a end-user machine where the user is changing their enviroment much more often is different. Even a lot of software requires kernel extensions on Mac OS X, and under Linux that software would be incredibly awkward. Linux is fast because it takes short cuts that Mac OS X doesn't, but these short cuts make Linux less usable. There is nothing wrong with having a faster kernel, but Linux's fast kernel comes at a cost.

[CharlesS]

I think his point was that OS X has the kernel extension feature in the first place. Yes, OS X has wireless drivers built-in, but if it didn't, all you'd need to do would be to drop a kernel extension in /System/Library/Extensions (although - can someone tell me why there's no equivalent in the /Library folder?) and be done with it. You wouldn't need to recompile the kernel...

[goMac]

Originally Posted by CharlesS

(although - can someone tell me why there's no equivalent in the /Library folder?)

I think that's for security reasons. An application running under an admin account could install kernel extensions without authorization if there was a /Library/Extensions.

[besson3c]

Originally Posted by goMac

Yes it does. Because if Linux was designed like OS X, it wouldn't matter what distro you're running. Just install the drivers for your wireless card, and you're done.

Then your complaint is with the Linux operating system as a whole (generalizing across all distros) and not with the kernel.

Using Linux on a server that is not going to be changing very much is one thing, but using Linux on a end-user machine where the user is changing their enviroment much more often is different. Even a lot of software requires kernel extensions on Mac OS X, and under Linux that software would be incredibly awkward. Linux is fast because it takes short cuts that Mac OS X doesn't, but these short cuts make Linux less usable. There is nothing wrong with having a faster kernel, but Linux's fast kernel comes at a cost.

Not short cuts. Linux leaves policy up to the user, it only focuses on the mechanism. It forces users to be very particular about what support they want included in their kernel. The result is a more optimized kernel. Comparing the performance of the OS X kernel vs. the Linux kernel would be difficult as there would be several variables to account for, but the Linux kernel is chosen for a number of different devices because it is so flexible and performs so well (and is free).

[besson3c]

Originally Posted by CharlesS

I think his point was that OS X has the kernel extension feature in the first place. Yes, OS X has wireless drivers built-in, but if it didn't, all you'd need to do would be to drop a kernel extension in /System/Library/Extensions (although - can someone tell me why there's no equivalent in the /Library folder?) and be done with it. You wouldn't need to recompile the kernel...

That gets into the differences between Mach and Monolithic kernels, which is an extensive argument. Both kernel designs have their strengths and weaknesses.

[- - e r i k - -]

Would that weakness be usability for mortals maybe?

[besson3c]

Originally Posted by - - e r i k - -

Would that weakness be usability for mortals maybe?

Maybe, although you could simply find a distribution that included support for absolutely every piece of hardware that can possibly be run under the OS, which is probably more like what Windows does.

Whether you run Mach or Monolithic, there is still a selection process of what support to include, and still a footprint that unneeded hardware support takes up which makes your kernel larger and less optimized.

The difference is simply in the ability to add and remove hardware support (i.e. kernel extensions) independently from the kernel as a whole, AFAIK. For OS X, this is perfect. For Linux users, many prefer the option of being able to run highly optimized kernels.

If you don't want a highly optimized Linux kernel, this in theory should simply be about the distro you decide to go with. I'm sure Linspire and Ubuntu and similar distros are very liberal about what hardware support is included with the default install.

[seanc]

Originally Posted by mitchell_pgh

Slashdot | Vista Licenses Limit OS Transfers, Ban VM Use

This is going to upset some of the geeks out there...

I'm guessing they will change this, but who knows...

Read the link that article on Slash came from. It's been updated. The licensing appears to be the same as XPs as in you can move it as many times as you like.

[Update] Vista Licenses Limit OS Transfers - Software Technology News by TechWeb

[mitchell_pgh]

Originally Posted by seanc

Read the link that article on Slash came from. It's been updated. The licensing appears to be the same as XPs as in you can move it as many times as you like.

[Update] Vista Licenses Limit OS Transfers - Software Technology News by TechWeb

Ugh... Slashdot has really gone down hill. I can't count the number of erroneous articles they have posted in recent months.

Sorry, nothing to see, moving along...

[goMac]

Originally Posted by besson3c

Then your complaint is with the Linux operating system as a whole (generalizing across all distros) and not with the kernel.

Ummmm. No it's not. I'm saying the Linux kernel wasn't designed for general users. Yes, any problem with the Linux kernel is going to carry over into all Linux operating systems, but I'm specifically saying the kernel is the problem.

[/QUOTE]Not short cuts. Linux leaves policy up to the user, it only focuses on the mechanism. It forces users to be very particular about what support they want included in their kernel.[/QUOTE]

Users don't want to have to worry about that. I don't want to take the time to maintain that. Normal users want plug and play. They don't want to sit around choosing kernel modules. Again, this may be great for your super power users, but this isn't what normal people want.

Originally Posted by besson3c

The result is a more optimized kernel. Comparing the performance of the OS X kernel vs. the Linux kernel would be difficult as there would be several variables to account for, but the Linux kernel is chosen for a number of different devices because it is so flexible and performs so well (and is free).

Which is great if you want a fast server or a embedded device. But this doesn't fall into the general use category. Read up on the "There is no free lunch theorem" on Wikipedia. It states that on average any computer algorithm will be no faster than any other. And I think this certainly applies to operating systems. In specific cases, Linux will be faster than Mach, and vise versa, but on average they'll even out and be just as fast as each other. Linux's strength is in speed. Mach's strength is in user friendlies. And user friendliness does translate to a speed gain, because it takes the user less time to perform a task.

Linux is not going to take over the mass market because it's developers do not understand that normal users do not want to worry about downloading packages, maintaining the kernel, or hand picking kernel modules. They just want their computer to work with whatever hardware and software they buy without hassle. This is where Mac OS X excels and Linux falls flat on it's face, and this is specially because of how the Linux kernel was designed among other things.

[besson3c]

[QUOTE=goMac]Ummmm. No it's not. I'm saying the Linux kernel wasn't designed for general users. Yes, any problem with the Linux kernel is going to carry over into all Linux operating systems, but I'm specifically saying the kernel is the problem.

Not short cuts. Linux leaves policy up to the user, it only focuses on mechanism. It forces users to be very particular about what support they want included in their kernel.

Users don't want to have to worry about that. I don't want to take the time to maintain that. Normal users want plug and play. They don't want to sit around choosing kernel modules. Again, this may be great for your super power users, but this isn't what normal people want.

Which is great if you want a fast server or a embedded device. But this doesn't fall into the general use category. Read up on the "There is no free lunch theorem" on Wikipedia. It states that on average any computer algorithm will be no faster than any other. And I think this certainly applies to operating systems. In specific cases, Linux will be faster than Mach, and vise versa, but on average they'll even out and be just as fast as each other. Linux's strength is in speed. Mach's strength is in user friendlies. And user friendliness does translate to a speed gain, because it takes the user less time to perform a task.

Linux is not going to take over the mass market because it's developers do not understand that normal users do not want to worry about downloading packages, maintaining the kernel, or hand picking kernel modules. They just want their computer to work with whatever hardware and software they buy without hassle. This is where Mac OS X excels and Linux falls flat on it's face, and this is specially because of how the Linux kernel was designed among other things.

Don't agree. Tell me what the difference is between bundling a kernel with support for all possible hardware enabled, and bundling a kernel with kernel extensions for additional hardware support?

The difference is that if a user wants to install support for chip sets and hardware families that are not supported under the OS, they'll have an easier time doing this under Mach. Now, how many OS X users do you know have done this?

Your argument also only applies to source-based package management for Linux. If you want to upgrade your kernel in Redhat, for instance, you just grab the RPM from up2date, install, and reboot...

GoMac, my only qualms with the stuff you write about Linux is your occasional inaccuracies, and this sort of anti-Linux bent you seem to have, for some reason. The truth is, Linux is a technologically sound operating system. What is holding it back in the Desktop space is simply policy, implementation, and a business plan. This is what has held back Unix for years, open source developers in general are not interested in developing business models and putting on suits. OS X could have very easily been built upon a Linux-based operating system (had NextStep already been Monolithic kernel based), and the end product wouldn't be any weaker than it is today.

I expect that Linux will eventually become more and more common for government and business as time progresses. There is no reason why people need to shell out money for something as mundane as Microsoft Office, when it is becoming easier and easier to do this on a Linux build, and in doing so save lots of money.

There is a hell of a lot of innovation coming out of the Open Source community, particularly in the server space. If you are at all interested in Microsoft becoming less and less relevant in the IT world, you should be silently cheering on Open Source, because it will be Open Source that pushes MS off its thrown, if any, not Apple.

[Doofy]

Does Mac OS design trail behind Mac hardware? Don't know. Just be glad you ain't using Windows.

[goMac]

Originally Posted by besson3c

Don't agree. Tell me what the difference is between bundling a kernel with support for all possible hardware enabled, and bundling a kernel with kernel extensions for additional hardware support?

If you want to add hardware that come out after your OS, you have to recompile your kernel. It also leaves you at the mercy of the Linux developers to determine what chipsets you can use, whereas under OS X the chipset vendors can ship their own drivers.

Originally Posted by besson3c

The difference is that if a user wants to install support for chip sets and hardware families that are not supported under the OS, they'll have an easier time doing this under Mach. Now, how many OS X users do you know have done this?

Hah. Many users do this. A lot of software ships with kernel drivers and users don't even realize it.

Originally Posted by besson3c

Your argument also only applies to source-based package management for Linux. If you want to upgrade your kernel in Redhat, for instance, you just grab the RPM from up2date, install, and reboot...

I can't imagine a normal user even doing that. Heck, I know a lot of users who don't even run software update.

Originally Posted by besson3c

GoMac, my only qualms with the stuff you write about Linux is your occasional inaccuracies, and this sort of anti-Linux bent you seem to have, for some reason.

I'm not anti Linux. I had a few Linux servers myself. I'm just realistic. Linux is good for some thigns, OS X others.

Originally Posted by besson3c

The truth is, Linux is a technologically sound operating system.

I'm not disagreeing. It's very stable.

Originally Posted by besson3c

What is holding it back in the Desktop space is simply policy, implementation, and a business plan.

No, the problem is that Linux is an operating system by geeks built for geeks. There is nothing wrong with that. It's just the Linux is not really normal user friendly. Your idea that a normal user is going to be interacting with RPM's shows the disconnect that Linux users have with how normal users use computers.

Originally Posted by besson3c

This is what has held back Unix for years, open source developers in general are not interested in developing business models and putting on suits. OS X could have very easily been built upon a Linux-based operating system (had NextStep already been Monolithic kernel based), and the end product wouldn't be any weaker than it is today.

It would have been weaker because you couldn't simply install drivers for new hardware in the operating system. You'd have to wait for Apple to support your hardware and then ship an entire software update just to support a new wireless card/wired card/usb card/firewire card. Under OS X as it is, a new card comes with a driver you install and you're good to go. This is very user friendly.

Originally Posted by besson3c

I expect that Linux will eventually become more and more common for government and business as time progresses. There is no reason why people need to shell out money for something as mundane as Microsoft Office, when it is becoming easier and easier to do this on a Linux build, and in doing so save lots of money.

I don't think OpenOffice's fate is tied to Linux. Microsoft Office and Linux are two separate issues. And Linux is honestly a pain to maintain on end user desktops. Can you imagine having to update someone's entire OS just to support a new network card? What a pain. This works great for servers whose hardware never changes. But not on an end user desktop machine.

Edit: You know, Linux users are always saying that Linux is about to take the OS market by storm. But it never does, and Mac is gaining market share at a rate well above Linux. There is nothing wrong with Linux, and there is nothing wrong with Linux users loving their OS, but seriously, we just aren't seeing Linux gain the ground we're always told it will.

Originally Posted by besson3c

There is a hell of a lot of innovation coming out of the Open Source community, particularly in the server space. If you are at all interested in Microsoft becoming less and less relevant in the IT world, you should be silently cheering on Open Source, because it will be Open Source that pushes MS off its thrown, if any, not Apple.

OpenOffice is again a separate program than Linux. And just because software is free that doesn't mean it's good for everyone.

Mach is also open source if that counts for anything.

[besson3c]

Originally Posted by goMac

If you want to add hardware that come out after your OS, you have to recompile your kernel. It also leaves you at the mercy of the Linux developers to determine what chipsets you can use, whereas under OS X the chipset vendors can ship their own drivers.

If by drivers you mean kernel extensions (which would be accurate), how many OS X users do you know of that have ever installed a kernel extension? Many would simply wait for the next OS revision.

Your argument is better taken with regards to the out-of-box experience. Again, the decision a distro makes about the default hardware support in a kernel does not represent the quality of the Linux kernel.

If any, one could argue that there is more hardware support under Linux, so therefore the Linux kernel is better, if a wide range of hardware support is one of your definitions of "better".

Hah. Many users do this. A lot of software ships with kernel drivers and users don't even realize it.

Such as? I remember a Cisco VPN extension, but that's about it... perhaps your experience differs from mine?

No, the problem is that Linux is an operating system by geeks built for geeks. There is nothing wrong with that. It's just the Linux is not really normal user friendly. Your idea that a normal user is going to be interacting with RPM's shows the disconnect that Linux users have with how normal users use computers.

Whether or not there are bugs or confusing usability issues with the Red Hat GUI interface I can't argue because I don't have experience there, but there is nothing terribly complicated about a pre-compiled/binary software installer, which is all RPM is. In my experience using up2date over the command line, it was pretty much just "GO" as far as the interaction needed for an update to take place. Whether or not you agree that the GUI is as user friendly as Apple's software update, it really can't be that far off, can it? This is not a complicated tool.

It would have been weaker because you couldn't simply install drivers for new hardware in the operating system. You'd have to wait for Apple to support your hardware and then ship an entire software update just to support a new wireless card/wired card/usb card/firewire card. Under OS X as it is, a new card comes with a driver you install and you're good to go. This is very user friendly.

Because the hardware is very predictable. How does this work in Windows? Probably similar to what could easily be done under Linux, if the "wait for the next OS update" concept doesn't apply (which it may very well, since several Linux distro updates have been occurring more frequently than Windows updates).

I don't think OpenOffice's fate is tied to Linux. Microsoft Office and Linux are two separate issues. And Linux is honestly a pain to maintain on end user desktops. Can you imagine having to update someone's entire OS just to support a new network card? What a pain. This works great for servers whose hardware never changes. But not on an end user desktop machine.

How often are there major changes in chipsets in something like a network card that would require a kernel level change? I think you are blowing this out of proportion...

Office and Linux may be separate issues, but if you don't intend to run Office, isn't there far less reason to run Windows?

Edit: You know, Linux users are always saying that Linux is about to take the OS market by storm. But it never does, and Mac is gaining market share at a rate well above Linux. There is nothing wrong with Linux, and there is nothing wrong with Linux users loving their OS, but seriously, we just aren't seeing Linux gain the ground we're always told it will.

How could this ever be measured? There is a lot of Linux and OSS usage in non-American countries, particularly in government. There is a lot of Linux usage in embedded devices (I'm assuming you are only referring to the Desktop market though). Do you count Linux-based appliances as part of the user-base? The Linux market is fragmented, and I don't think there is any reliable way to measure it as a whole (except maybe on publicly accessible machines such as web servers).

OS X is defeating Linux handily in the Desktop space, that is for sure. I should have qualified what I said by saying that when Linux does start to replace Microsoft installations, it will most likely begin in government and business, not home Desktop computers.

To me it just seems like plain and simple logic. How can you compete with free? Once it becomes viable in the hearts and minds of people making decisions, as long as there is a compelling reason to switch (i.e. cost savings), what would be stopping people?

[- - e r i k - -]

You can always compete with free. Free is not a desirable quality for most people. The perception is the more you pay for it the more it's worth.

Hard to get your mind around, I know.

[goMac]

Originally Posted by besson3c

If by drivers you mean kernel extensions (which would be accurate), how many OS X users do you know of that have ever installed a kernel extension? Many would simply wait for the next OS revision.

As a system administrator, this is unthinkable. We're not going to upgrade someone's OS just to get EyeTV working (We just bought hundreds of copies of EyeTV). It is unacceptable that a user who goes out and buys something like a Pre-N card cannot get it to run under Linux because the kernel does not support these devices, when under OS X they can just install drivers.

Originally Posted by besson3c

Your argument is better taken with regards to the out-of-box experience. Again, the decision a distro makes about the default hardware support in a kernel does not represent the quality of the Linux kernel.

Of course it does. The design of the Linux kernel makes it so that the user is dependent on the distributions maintainers for which hardware they can and cannot run. Linux users complain all the time that : blank : hardware doesn't support Linux, when it's Linux's own fault because any support has to be compiled into the kernel making it very complex.

There are a lot of power users who want to install hardware, drivers, and have it work. We don't want to be locked in by our distribution into certain hardware, or have to recompile the kernel. Linux does not just work. (TM)

Originally Posted by besson3c

If any, one could argue that there is more hardware support under Linux, so therefore the Linux kernel is better, if a wide range of hardware support is one of your definitions of "better".

There isn't more hardware support under Linux, but ok.

Originally Posted by besson3c

Such as? I remember a Cisco VPN extension, but that's about it... perhaps your experience differs from mine?

EyeTV for one. Virtual PC. A lot of hardware devices. The list goes on, but it's besides the point.

Originally Posted by besson3c

Whether or not there are bugs or confusing usability issues with the Red Hat GUI interface I can't argue because I don't have experience there, but there is nothing terribly complicated about a pre-compiled/binary software installer, which is all RPM is. In my experience using up2date over the command line, it was pretty much just "GO" as far as the interaction needed for an update to take place. Whether or not you agree that the GUI is as user friendly as Apple's software update, it really can't be that far off, can it? This is not a complicated tool.

Again, Linux users don't seem to understand the problem. Users don't want to run software update. They don't want to have to recompile the kernel. They don't want to have to install a new one. They don't want to know what a kernel even is. This isn't a GUI issue at all.

Originally Posted by besson3c

Because the hardware is very predictable. How does this work in Windows? Probably similar to what could easily be done under Linux, if the "wait for the next OS update" concept doesn't apply (which it may very well, since several Linux distro updates have been occurring more frequently than Windows updates).

I don't need Windows updates to install new hardware on my Windows machine. I don't need Mac OS X updates to install new hardware on my Mac. Why do I need to update Linux to install new hardware under Linux? There are new chipsets that come out all the time. Linux has always lagged behind. Because vendor's cannot provide their own drivers, it means hardware has to be reverse engineered for Linux, which takes quite a while.

Originally Posted by besson3c

How often are there major changes in chipsets in something like a network card that would require a kernel level change? I think you are blowing this out of proportion...

Often enough. Linux went a very long time without support for Centrino wireless.

Originally Posted by besson3c

Office and Linux may be separate issues, but if you don't intend to run Office, isn't there far less reason to run Windows?

Nope. There is still a lot of other Windows software. Not to mention the scenarios you're talking about with Linux would be unacceptable under a lot of production environments. If we have to spend time upgrading someone's operating system to get their wireless card working, we'll start spending our time with a different operating system, thank you very much.

Originally Posted by besson3c

How could this ever be measured? There is a lot of Linux and OSS usage in non-American countries, particularly in government.

A few countries.

Originally Posted by besson3c

There is a lot of Linux usage in embedded devices (I'm assuming you are only referring to the Desktop market though). Do you count Linux-based appliances as part of the user-base? The Linux market is fragmented, and I don't think there is any reliable way to measure it as a whole (except maybe on publicly accessible machines such as web servers).

Here is the thing about Linux. Linux is very good when it's being deployed on a device which in which it's known that the device is only going to be doing a few things. This is because Linux is optimized for very specific purposes. When it's deployed for general use, it falls flat on it's face because the optimizations that have been made to it result in usability issues in other area. The kernel issues that I'm talking about is one example.

Originally Posted by besson3c

OS X is defeating Linux handily in the Desktop space, that is for sure. I should have qualified what I said by saying that when Linux does start to replace Microsoft installations, it will most likely begin in government and business, not home Desktop computers.

I made a push at work once for replacing our Windows machines with Linux. (We're about %10 Windows). So I sat down, tried to get Linux working to replace the tasks that our Window's machines did, and that was when I realized that maintaining Linux was a pain, and simply getting it set up was a pain. Again, I don't care how fast or geeky Windows is. In workplaces, I want the machine to take 10 minutes to set up. When it takes 10 minutes to administer touch screen drivers on a Windows machine, and I simply can't get them working under Linux, Windows wins. I don't care how free or open source Linux is.

Originally Posted by besson3c

To me it just seems like plain and simple logic. How can you compete with free? Once it becomes viable in the hearts and minds of people making decisions, as long as there is a compelling reason to switch (i.e. cost savings), what would be stopping people?

By your logic pico should have driven Word out of the market long ago.

[goMac]

Let me expand a bit on Microsoft Office vs. Open Office. We bought 3000 laptops this year (Macs)for students. When time came to build an image, we sat down and compared potential candidates for what office software we could use. The choices were:

a) OpenOffice
b) NeoOffice
c) Microsoft Office

Do you know what we chose? Microsoft Office. Why? Because we knew it would be compatible with other word files. We knew it worked. It had a very good interface. It was well polished. It was very Mac like. It was just all around better.

Do we like Microsoft? No. Are we worried about Macro viruses? Yes. Are we paying a bunch of money to Microsoft? Yes. But just because a piece of software is free doesn't mean it cuts it. Microsoft Office is the best solution on the table.

[itai195]

Originally Posted by besson3c

To me it just seems like plain and simple logic. How can you compete with free? Once it becomes viable in the hearts and minds of people making decisions, as long as there is a compelling reason to switch (i.e. cost savings), what would be stopping people?

In theory that's true, but the cost of an OS license is a relatively insignificant portion of the TCO calculations most corporations do. That said, companies (like mine) seem to have committed to the TCO calculations they did in the late 90s. They're in bed with Microsoft to such an extent that a change on the desktop now would be massively expensive.

[CharlesS]

Originally Posted by goMac

I think that's for security reasons. An application running under an admin account could install kernel extensions without authorization if there was a /Library/Extensions.

Well, StartupItems is in Library, and everything in there runs as root. The OS just makes sure the startup items are owned by root, and doesn't run them otherwise. Why is that good enough for startup items (which can cause a load of problems if they are malicious and running as root), but not kernel extensions?

Originally Posted by besson3c

Don't agree. Tell me what the difference is between bundling a kernel with support for all possible hardware enabled, and bundling a kernel with kernel extensions for additional hardware support?

Umm... the fact that the former is impossible?

The difference is that if a user wants to install support for chip sets and hardware families that are not supported under the OS, they'll have an easier time doing this under Mach. Now, how many OS X users do you know have done this?

Well, lessee: my machine is running the following third-party kernel extensions:

AmbrosiaAudioSupport.kext - for a demo version of WireTap Pro that I tried out a while ago. And since WireTap needs a kext, Snapz Pro probably would, too.
USBOverdrive.kext - a very useful driver for sundry USB devices
HotSync Classic Seize.kext - for Palm Desktop; PDAs being less common than they used to be, but not unheard of
LittleSnitch.kext - for a demo of Little Snitch that I tried out a while ago - I decided I didn't need it, but a lot of people swear by this thing

Shameless plug: My app, Pacifist, has a Kernel Extension Report feature that will scan your Extensions folder and let you know what third-party kernel extensions YOU have installed.

edit: oh, and IIRC, the drivers for the keyboards and mice from Logitech and Microsoft include kernel extensions. I'd say those third-party meeses are quite popular, no? Sure, you can get by without the drivers for the mice, but you really do want them for the keyboards, or else the Command and Option keys end up being the reverse of what they should be.

edit again: oh, and I forgot to mention that I also have a few kernel extensions installed for CHUD, which is part of the developer tools. These are Apple-provided, but didn't come with the OS.

[Chuckit]

Originally Posted by CharlesS

Well, StartupItems is in Library, and everything in there runs as root. The OS just makes sure the startup items are owned by root, and doesn't run them otherwise. Why is that good enough for startup items (which can cause a load of problems if they are malicious and running as root), but not kernel extensions?

The reason is not actually security, as I understand it. The reason is that, unlike the rest of the things in Library folders, kernel extensions are loaded by BootX. The kernel team decided that making it look in several domains for extensions would add undesirable complexity to BootX for little benefit. One of Apple's engineers posted on the subject a couple of years back.

[The Godfather]

Originally Posted by besson3c

To me it just seems like plain and simple logic. How can you compete with free? Once it becomes viable in the hearts and minds of people making decisions, as long as there is a compelling reason to switch (i.e. cost savings), what would be stopping people?

Sleeping with a beautiful hooker for money, vs sleeping with a not-so-hot hooker for free.

[besson3c]

Let's backpeddle here. It seems like we are talking past each other, and perhaps there are some things you don't understand (unless you are just misspeaking)

1) Kernel support is only necessary for *classes* of hardware. For instance, there is USB and Firewire support options. There is support for chipsets and families of hardware (e.g. Nvideo and AMD cards)

2) You are free to download and install drivers to hardware. Individual peripherals and stuff often do not require kernel level support. You don't need to recompile to add kernel level support specifically for each and every piece of hardware you have. Kernel level support is one of the lowest levels of hardware communication. Many pieces of hardware are negotiated by a more general driver/bridge. The kernel handles some low level arbitration, not providing drivers for everything under the sun.

3) You do not need to recompile your kernel to add or remove hardware. You'd need to recompile your kernel if you decide you want to replace your PCI bus with some other form of expansion, you wanted to connect a machine to your SAN, etc. and your kernel did not already support this. If you want absolute plug-and-play, don't choose a distro that expects you to compile and optimized kernel from source.

Originally Posted by goMac

As a system administrator, this is unthinkable. We're not going to upgrade someone's OS just to get EyeTV working (We just bought hundreds of copies of EyeTV). It is unacceptable that a user who goes out and buys something like a Pre-N card cannot get it to run under Linux because the kernel does not support these devices, when under OS X they can just install drivers.

Under Linux if you bought EyeTV, you could simply install a device driver for something like this. Kernel support would only need to be enabled for Firewire, assuming that EyeTV is a firewire based device (looked into it a while ago, thought it was expensive, passed on getting it for myself, don't remember too much about it).

Whatever Virtual PC requires could surely be provided in the form of a driver. Be careful about your usage of "driver" vs. "kernel extension", they are not the same thing.

Of course it does. The design of the Linux kernel makes it so that the user is dependent on the distributions maintainers for which hardware they can and cannot run. Linux users complain all the time that : blank : hardware doesn't support Linux, when it's Linux's own fault because any support has to be compiled into the kernel making it very complex.

By now, hopefully I've addressed this clearly.

There are a lot of power users who want to install hardware, drivers, and have it work. We don't want to be locked in by our distribution into certain hardware, or have to recompile the kernel. Linux does not just work. (TM)

Then they should choose a distro that is extremely liberal about enabling these options. This will provide them with as much of a plug-and-play experience as is possible as far as accessing what is supported under Linux.

There isn't more hardware support under Linux, but ok.

Than what, OS X? Sure there is... Why did you pick EyeTV? Because it's one of the only players available for the Mac. What about the entire Hauppauge line? What about expansion options such as RAIDs, SANs, SCSI based? There is a ton of hardware support missing in OS X, primarily non-commodity type hardware that you can't just drive down to Best Buy to get, granted.

Again, Linux users don't seem to understand the problem. Users don't want to run software update. They don't want to have to recompile the kernel. They don't want to have to install a new one. They don't want to know what a kernel even is. This isn't a GUI issue at all.

Well, a user of any OS is going to have to make some sort of update to support a *class* of hardware that didn't exist when they installed the OS, just the same as any other user. I guess this relates to your thinking that Linux users need to recompile their kernel for each and every driver they want to use.

I don't need Windows updates to install new hardware on my Windows machine. I don't need Mac OS X updates to install new hardware on my Mac. Why do I need to update Linux to install new hardware under Linux? There are new chipsets that come out all the time. Linux has always lagged behind. Because vendor's cannot provide their own drivers, it means hardware has to be reverse engineered for Linux, which takes quite a while.

If hardware support lags behind Windows or Mac OS it is due to a business reality, not due to the merits of the Linux kernel. I'd say that in my cases the Mac is in the exact same boat. Look at all the video cards that are specially made to run under Windows, for instance.

Nope. There is still a lot of other Windows software. Not to mention the scenarios you're talking about with Linux would be unacceptable under a lot of production environments. If we have to spend time upgrading someone's operating system to get their wireless card working, we'll start spending our time with a different operating system, thank you very much.

You'd only have to do this if they installed wireless <whatever the next protocol after G will be> and this support wasn't available at all, or built in to the kernel.

All I'm saying is that you are blowing this out of proportion for the purpose of trying to make your argument.

A few countries.

Several, actually. The US government is really lagging behind.

I made a push at work once for replacing our Windows machines with Linux. (We're about %10 Windows). So I sat down, tried to get Linux working to replace the tasks that our Window's machines did, and that was when I realized that maintaining Linux was a pain, and simply getting it set up was a pain. Again, I don't care how fast or geeky Windows is. In workplaces, I want the machine to take 10 minutes to set up. When it takes 10 minutes to administer touch screen drivers on a Windows machine, and I simply can't get them working under Linux, Windows wins. I don't care how free or open source Linux is.

This could be due to several factors and variables which you haven't accounted for in this paragraph...

By your logic pico should have driven Word out of the market long ago.

No, that's why I said "the hearts and *minds*" of people. It doesn't make logical sense to switch to something lacking features that are needed.

If there was a product that did everything that the current product in use did, i.e. had all the features needed, worked well, was easy to install and get setup and running, etc. why would a company *not* make this switch if it represented cost savings? I'm saying the cost savings would be a primarily motivator to the guys in suits, but I'm not saying that it would be the only one.

[besson3c]

Originally Posted by CharlesS

Umm... the fact that the former is impossible?

Not under the BSDs it isn't...

OS X's kernel is simply a general purpose kernel with support for hardware that Apple would expect you'd want to use. If you need to run hardware that requires kernel level arbitration that isn't built in to the OS, this is where a kernel ext would come in... this is not to be confused with a driver. All of the stuff you listed are just drivers. In OS X, it seems like there is not really a distinction between kernel exts and drivers?

For USB stuff would simply need the USB support built in, and Little Snitch may need something like the iptables connip (or whatever it is called) to establish ipfw rules based on IP addresses. The latter may very well be included by default in a number of distros.

The point is, there is no "support Little Snitch" type options in the kernel. I think what you guys have an issue with is the fact that Linux creates a lot of guesswork such as "does this distro include a kernel with support for a lot of different kinds of hardware, or was it very conservative in enabling hardware support in the kernel build options", as well as all of the general "Linux is hard!" frustrations. This is simply one of the tradeoffs of one of the key design decisions of Linux. My point with regards to this is simply that the Linux kernel *itself* is fine. What you have grievances with is the staggering amount of choices and flexibility in Linux and the lack of an Apple like vendor that would match the sort of usability and friendliness you expect in a Desktop OS.

[besson3c]

Originally Posted by itai195

In theory that's true, but the cost of an OS license is a relatively insignificant portion of the TCO calculations most corporations do. That said, companies (like mine) seem to have committed to the TCO calculations they did in the late 90s. They're in bed with Microsoft to such an extent that a change on the desktop now would be massively expensive.

True, one would have to account for staffing and such. However, in certain environments a transition like this could be slowly phased in. Our local school system is using Linux machines with a government pilot/grant, and may slowly phase in Linux machines to other labs too, since they've come up with a pretty solid build.