Mac OS X 10.4 (Tiger) Coming for the PC?

[MagnusDredd]

UNIX not UNIX

Basically my take on the whole UNIX not UNIX thing, is that as long as my scripts work, I don't care too much. This of course means that I don't write things that rely on /proc (Linux), /dev/ad (BSD), or things like that.

Since I don't write device drivers (which are different on Solaris and AIX), I might be interested in kernels intellectually but not seriously worried where the kernel was derived from. I'm happy as long as it's robust, stable, and secure. I'm happier if it's quick. But in fact, it's the userland that I care about. Do I have sed, awk, echo, etc... If something is missing like "watch" on OSX, and included on AIX (but does something different) I try to find a port... BTW "watch" is part of procps http://procps.sourceforge.net/ which contains a ton of utilities that only work on linux as far as I know.

I took a class that used a frame running Digital UNIX 4.0b. Older than hell. painful to use.

 

[realtime]

Why focus on MacOS X? Apple makes a lot of software these days. What if they shipped some of their iLife stuff on Windows using this product as a quicker way to port code?

And as for MacOS X running on an Intel processor, I don't know why people assume that means "generic PC hardware." A processor is a processor, and unfortuantely x86 processors have consistently stayed ahead of PowerPC over the past few years. If Apple's OS still only ran on Apple hardware, but you had an Intel or AMD processor under the hood, hopefully you'd never know the difference (except your desktop Mac would be faster and your laptop would use less power.) This wouldn't even imply that your new IntelMac could run Windows, or even linux.

Mac OS X has always been able to run on X86 processors, and Apple maintains that compatibility, not necessarily to allow the OS to be used on other platforms, but to ensure that little-endian processors (such as x86) will always be an option for Apple hardware. If you browse ANY version of Mac OS X, you'll find plenty of evidence in the private classes, as well as in some of the public API. FWIW, the kernel that ships with Mac OS X is not the same as you are able to build from the Darwin open source; among other things, Apple adds some lightweight measures to ensure it is running on Apple hardware, disguised as "machine" optimizations, firmware, etc.

 

[~loserman~]

Put the Crack Pipe Down

If Apple's OS still only ran on Apple hardware, but you had an Intel or AMD processor under the hood, hopefully you'd never know the difference (except your desktop Mac would be faster and your laptop would use less power.)

I own and use 5 brand new shiny PC laptops and none of the can hold a candle to the battery life of my powerbook.

 

[bigandy]

but apple don't want to die. so it's not gonna happen.

if they did it, mac software development stops, macs stop selling, they lose all their money and FLUSH down the pan.

and we'd all be mighty angry at apple for dying because = no more decent computers.

 

[SPUY767]

I like cheese.

i hope this doesn't happen.. i mean if it does.. whats pc' users incentive to buy macs if they can just get the great software and use it on their blazing fast alienwares and falcon norths?

A dual 2.5 GHz g5, considerably less than a comparable alienware or falcon, or any other low volume PC maker, with concern to price, will eat it alive. Absolutely devour it, I'm talking Trials of life, orcas eating seals here. In terms of pure power, Number crunching, (that is what a computer is designed for, am I wrong), a PPC970 destroys the x86 in a 32 or 64 bit incarnation. Given the difference in games, i.e. not being optimized for PPC, but being optimized for Windows, game Benchmarks are trash. However, if you take real spplications like photoshop, Cinema 4D, Maya, After Effects. Apps that are all written to make the most effective use of a computers available resources, no windows running machine out there is faster. I'm sorry man, you're wrong.

 

[BWhaler]

Worst. Rumor. Ever.

 

[Lacero]

Not. Worst. Rumor. EVER. If you want to scare Microsoft ****less, porting OSX to the PC would be the ticket. And if OSX could run Windows app natively or by way of hardware emulation, then Microsoft be ****less.

 

[weezer51402]

Anybody remember OPENSTEP? Rhapsody DR1 was essentially OPENSTEP with a mac-ish theme. They had to port that to PPC along with all of the OPENSTEP APIS which look almost identical to cocoa. Yes OS X isn't rhapsody I know but the kernel it very similar, it still runs on top of mach and the API is very very similar. I'm sure if apple had intentions of selling os x for x86 they wouldn't need this software which will only create a licensing problem that will cut their margins. It's fair to assume since the API was ported from m68k and then to intel that apple probably maintained the x86 port. They clearly maintain kernel updates with the darwin project and it's not like they need to rewrite the API each time theres a new release they just update it which shouldn't be to hard for them. Furthermore the problem of compiling for both platforms could be solved using the old FAT system employed by OPENSTEP that allowed for one neat package that could run on m68k, sparc, intel and hppa. To address the windows compatibility issue they could simply write a nice wrapper for wine and with X11 it shouldn't be that difficult at all. I'm not sure apple would want to run wine on X86 systems because it would completely obsolete apples hardware...why buy ppc when you can run windows and mac applications natively on the same box? well thats my 2 cents

One last thing, the only place I can see this software proving usefull is in emulating carbon but for the long term licensing cost for apple they would probably opt to just port carbon for x86.

 

[MagnusDredd]

Anybody remember OPENSTEP? Rhapsody DR1 was essentially OPENSTEP with a mac-ish theme. They had to port that to PPC along with all of the OPENSTEP APIS which look almost identical to cocoa. Yes OS X isn't rhapsody I know but the kernel it very similar, it still runs on top of mach and the API is very very similar. I'm sure if apple had intentions of selling os x for x86 they wouldn't need this software which will only create a licensing problem that will cut their margins. It's fair to assume since the API was ported from m68k and then to intel that apple probably maintained the x86 port. They clearly maintain kernel updates with the darwin project and it's not like they need to rewrite the API each time theres a new release they just update it which shouldn't be to hard for them. Furthermore the problem of compiling for both platforms could be solved using the old FAT system employed by OPENSTEP that allowed for one neat package that could run on m68k, sparc, intel and hppa. To address the windows compatibility issue they could simply write a nice wrapper for wine and with X11 it shouldn't be that difficult at all. I'm not sure apple would want to run wine on X86 systems because it would completely obsolete apples hardware...why buy ppc when you can run windows and mac applications natively on the same box? well thats my 2 cents

One last thing, the only place I can see this software proving usefull is in emulating carbon but for the long term licensing cost for apple they would probably opt to just port carbon for x86.

OpenStep was simply a newer version of NeXTstep a few years later and with the ability to run on top of other OSes. Rhapsody was a version of OpenStep ported to the PPC architecture. Rhapsody/OSX Server 1.x was virtually identical to OpenStep 4.2, except I think for the display model. NeXT/OpenStep applications could be changed slightly to run on Rhapsody/OSX Server 1.x. This is why most of the early graphical OSX software was written by NeXT/OpenStep development houses. This is also why you still use NeXTStep API calls (Cocoa is simply a newer version of the NeXT API) like NSwindow (create a window) and NSbutton (create a button) have been in use since the OS was called NeXTStep. Here's a hint, most of the OSX Cocoa API calls that begin with "NS" come from NeXTStep...

Some useful links about the history of NeXTStep/OSX, which are simply versions of the same Operating System. OS News has a great article here. Kernel Thread has a brief and a not so brief history of OSX, covering the stuff from NeXT.

Even were carbon (the original macintosh API, "Macintosh Toolbox" minus about 10% of the calls) to be ported to x86 again (it has been at least partially done before), there would still have to be a CPU emulator for older applications. Applications make system calls as well as API calls. So while it is much faster to only emulate the CPU for the system calls, the way Apple did when they switched CPU architectures from m68k to PPC, it's still not the best case scenario by any means. The only reason that it worked as well as I hear that it did (apparently many users never noticed) is because the PPC is so much more efficient than the old m68k series processors. Currently PearPC, a PPC emulator for x86 is so slow that it takes 5 hours to install OSX on an Athlon 1600+. According to another person a Athlon64 3200+ runs OSX on PearPC about as fast as a Pentium 120 runs Windows 98. This is due to the fact that most x86 processors have much fewer registers than the G4. I would imagine however that if you were to optimize PearPC for the Athlon64 you could possibly get much better emulation speed due to the greater number of registers on the Athlon 64's. I'm fairly sure it would take a great deal of work though.

Note: amazingly enough, the m68k architecture is still in use today, just not in desktop computers... Look up "coldfire" on google... To see more on classic MacOS on Intel, search google for: "mac os" "star trek" intel...

 

[weezer51402]

Not exactly, if I understand what you mean by display model correctly. Yes they looked different but OS X server 1.0-1.2 used PostScript as well as the developer releases of rhapsody and NeXTSTEP/OPENSTEP. When they finally switched over to the OS X environment that we now know they used quartz. It is also correct the OPENSTEP API is almost identical to the cocoa api, their are just newer things in the cocoa API like NSDrawer for instance. As for carbon I *think* alot of those technologies were yielded from the taligent project and *maybe copland*. The idea at the time was to be able to run pink/taligent as apples next generation(as we all know the project died) but just like OS X it would need to run some classic apps and that is where I believe beginings to carbon began. As for OPENSTEP programmers they're still around. If your looking for some good software made by some of these people I would check out some stuff by Andrew Stone and also a nice FTP app called RBrowser. As for 68k boxes I plan on getting a quadra son to run A/UX so they are very useful. The coldfire project looks pretty cool.

 

[wrldwzrd89]

<snip>
Even were carbon (the original macintosh API, "Macintosh Toolbox" minus about 10% of the calls)
<snip>

Actually, "Carbon" isn't truly a subset of the original "Macintosh Toolbox". You are correct that about 10% of the functions in the Macintosh Toolbox are not in Carbon. However, some of those removed functions have been replaced with Mac OS X-compliant versions, and some altogether new calls have been added.

 

[bretm]

but apple don't want to die. so it's not gonna happen.

if they did it, mac software development stops, macs stop selling, they lose all their money and FLUSH down the pan.

and we'd all be mighty angry at apple for dying because = no more decent computers.

What's to stop Apple from being THE high end x86 computer maker, AND the number one software supplier.

Forget the OS, Apple could possibly make more money by giving in the hardware fight and porting all thier software to the Windows platform. They're practically there now with iTunes and iPods. If they start porting iPhoto, Final Cut Pro (you know how many PC Avid users would love to spend an extra grand and have FCP as well running on the same machine? - all of 'em.) , Motion, Soundtrack, pages, keynote, and whatever future competition they have for photoshop their potential market increases by 90%. To instantly have a 90% growth potential would be amazing.

At some point there will be an obvious line they have to cross. Let go the loyalists and increase their potential software market exponentially, or dig in even deeper to fight back against microsoft and try to expand their hardware market share. Since they have the best hardware and OS and are still losing market share, how's that gonna happen? That battle is lost. They may never cross the line until Jobs is gone - OR until they have a full software line available. Until they have a better app to compete with every microsoft app and every adobe app (since adobe is slowly abandoning the mac) they can't cross over. It will have to be a sort of grand surprise attack. "press release: Tomorrow all Apple apps will run natively on either Mac OS X or Windwows XP in the same package. Development of new mac hardware will end in 24 months. Current owners of apple software can upgrade to dual platform versions with a nominal fee...."

AND then Apple announces their new sub-$1000 laptops that run windows XP and look surprisingly like an aluminum powerbook. Their laptops are already competitively priced compared to windows notebooks. Windows users would jump at the chance to own an apple laptop running their beloved windows xp os.

And if I can come up with these goofy scenarios, go figure what the apple execs sit around and think about all day!

 

[JRM PowerPod]

I will see you for the release of Tiger for MAC (good computers) at WWDC in a few months, and will see you at the release for Longhorn (for bad computers) some time for my 80th birthday. BTW im 18.

Keep it real

 

[SPUY767]

I'm not sure apple would want to run wine on X86 systems because it would completely obsolete apples hardware...

Although buddy man's grammar is poor, I don't understand this constant thinking that Intel or AMD's hardware has a chance in hell of pushing IBM's into obsolescence. It is a different kind of processor folks. The AMD runs at a slower clock speed and outpaces Intel's hardware. IBM's chips run at a lower clock speed, and outpace them both in just about everything but games and fluked benchmarks. Does anyone on this board know the processor that powers the worlds fastest computer? A 400Mhz G4. IBM's hardware isn't going anywhere. Intel is having a whale of a time squeezing any more BS numbers out of the already burgeoning x86 architecture. Bottom line: you pit top of the line intel hardware against top of the line apple hardware, and run benchmarks on apps that are highly optomized for each processor, and you will find an ******* where there was none before. It will be the one that the 970 ripped in the P4.

 

[MagnusDredd]

APIs

Actually, "Carbon" isn't truly a subset of the original "Macintosh Toolbox". You are correct that about 10% of the functions in the Macintosh Toolbox are not in Carbon. However, some of those removed functions have been replaced with Mac OS X-compliant versions, and some altogether new calls have been added.

Right... the replacements were for some of the dodgy crap that was left over from the m68k days from what I understand. Basically carbon is a newer modernized version (major version upgrade if you will) of the Macintosh Toolbox, that has been ported to NeXTStep/OpenStep/OSX. The same way that you could say that Windows 2000 is an updated version of NT4. NT4 was so weak, that if you breathed on it hard, it would fall over. Win2k is far more robust. Took em 40 million lines of code added to the heap that was NT4 to make it work though.

This is the same way that you can use much of the Cocoa API in programming for Linux, while using the GnuStep API (at least the part directly based on the NeXT API). So NSWindow and NSButton can be used in GNUStep for making GUI elements under linux in this way.

The thing I find so funny about this, is the fact that while carbon is OSX native, it's not the true native API for OSX. Apple's native API for their flagship OS was created by a company other than Apple. Though many of the people responsible for NeXT are now at Apple, Avie Tevanian for instance. Speaking of which... I understand from rather reputable sources that the Apple's purchase of NeXT was something like a reverse takeover. The story goes that the leadership roles at Apple have been replaced by old NeXT guys.

 

[MagnusDredd]

Yes they looked different but OS X server 1.0-1.2 used PostScript as well as the developer releases of rhapsody and NeXTSTEP/OPENSTEP. When they finally switched over to the OS X environment that we now know they used quartz.

*nod* I was aware that OSX public beta was the beginnings of quartz. What I was unsure of was whether the display model had been tweaked in the progression from OpenStep to Rhapsody. It's obvious to anyone who has used both OSX Server 1.0-1.2 and OSX 10.x that they use completely different models. I was (still am) unsure of whether the display model was the exact same one in both OpenStep 4.2 and OSX Server 1.0... They look nearly identical, but... Furthermore I really didn't use NeXTStep except for a few occasions... While I do network/systems administration and building/repairing/configuring of networks, I'm not a serious programmer. Bash/Apple/VB(shudder) scripts, sure. Cocoa/Carbon/Win32/GTK+, not really... I have read a few technical books on Cocoa, the Macintosh Toolbox, and VB *bleah*.

It is also correct the OPENSTEP API is almost identical to the cocoa api, their are just newer things in the cocoa API like NSDrawer for instance.

That all follows logic, since the API has, and will continue to evolve. This is the same way that Cocoa has gained new calls with every OSX version upgrade.

The idea at the time was to be able to run pink/taligent as apples next generation(as we all know the project died) but just like OS X it would need to run some classic apps and that is where I believe beginings to carbon began.

That's not something I haven't heard about. However I do know that the reason that Carbon was a #1 project at Apple during the early OSX days, was because Adobe (and most other major Apple development houses) told Apple that there was no way they were going to re-write their possibly multiple million line applications for a new API. Not even if they did have a team that knew it.

 

[wrldwzrd89]

Right... the replacements were for some of the dodgy crap that was left over from the m68k days from what I understand. Basically carbon is a newer modernized version (major version upgrade if you will) of the Macintosh Toolbox, that has been ported to NeXTStep/OpenStep/OSX. The same way that you could say that Windows 2000 is an updated version of NT4. NT4 was so weak, that if you breathed on it hard, it would fall over. Win2k is far more robust. Took em 40 million lines of code added to the heap that was NT4 to make it work though.

This is the same way that you can use much of the Cocoa API in programming for Linux, while using the GnuStep API (at least the part directly based on the NeXT API). So NSWindow and NSButton can be used in GNUStep for making GUI elements under linux in this way.

The thing I find so funny about this, is the fact that while carbon is OSX native, it's not the true native API for OSX. Apple's native API for their flagship OS was created by a company other than Apple. Though many of the people responsible for NeXT are now at Apple, Avie Tevanian for instance. Speaking of which... I understand from rather reputable sources that the Apple's purchase of NeXT was something like a reverse takeover. The story goes that the leadership roles at Apple have been replaced by old NeXT guys.

You're absolutely right there - the way Cocoa came about as Apple's preferred API is strange to say the least. All the points you mentioned contributed to its creation and establishment as the default.

As far as NeXT goes, I think it was a good deal for both companies, regardless of how it was implemented. Avie Tevanian deserves to be at Apple, and seems to be thriving there.

 

[bazzler]

What if.....

Transitive tech. was incorporated into OSX and used to run x86 programs on mac ppc hardware? Apple could hardwire a list of banned programs into the OS (ie franchises that are already exist on ppc) so that software development does not dry up for ppc and windows users would have an extra incentive to switch. And im not considering games, i mean business programs etc that are only exist on x86 etc. As more people switch and use these programs on the mac, the incentive to write ppc native , optimised versions of software would increase and then people could be weaned off the x86 versions to mac native versions. Now this is just an idea, im not familiar with programming , hardware incompatibilities etc - so please could people refrain from the ' thats stupid because ...' type putdowns.

 

[wrldwzrd89]

Transitive tech. was incorporated into OSX and used to run x86 programs on mac ppc hardware? Apple could hardwire a list of banned programs into the OS (ie franchises that are already exist on ppc) so that software development does not dry up for ppc and windows users would have an extra incentive to switch. And im not considering games, i mean business programs etc that are only exist on x86 etc. As more people switch and use these programs on the mac, the incentive to write ppc native , optimised versions of software would increase and then people could be weaned off the x86 versions to mac native versions. Now this is just an idea, im not familiar with programming , hardware incompatibilities etc - so please could people refrain from the ' thats stupid because ...' type putdowns.

That's a great idea - but there's one little problem: how to update the ban list when it needs updating?

 

[weldon]

Transitive tech. was incorporated into OSX and used to run x86 programs on mac ppc hardware?

There's a lot more to running Windows programs under OS X than just using the Transitive technology to translate x86 instructions into PPC instructions. There's a whole set of shared libraries, Windows API calls, and assumptions about hardware that needs to be addressed.

 

[bazzler]

The banned list could be updated by Software Update, with unknown programs banned by default. As for hardware assumptions , windows api calls and other stuff that i dont understand; where theres a will theres a way.Maybe a deal with microsoft , as clearly this would screw virtual pc so they could be quite willing to have a small bit of revenue from every osx sale, than lose this revenue stream completely. But then again, they have gajillions in the bank and would cut off the nose to spite the face. perhaps. who knows, its just conjecture; although apple have a way of making cool stuff like this happen. In my wildest fanboy dreams, they unleash a new mac based on a modified cell processor that has a full altivec unit stitched on and core image addresses the secondary vector units to give amazing 3d performance and sony lets it be ps3-game compatible and steve jobs becomes grand high cheese of all computerdom.

 

[RacerX]

*nod* I was aware that OSX public beta was the beginnings of quartz. What I was unsure of was whether the display model had been tweaked in the progression from OpenStep to Rhapsody. It's obvious to anyone who has used both OSX Server 1.0-1.2 and OSX 10.x that they use completely different models. I was (still am) unsure of whether the display model was the exact same one in both OpenStep 4.2 and OSX Server 1.0... They look nearly identical, but... Furthermore I really didn't use NeXTStep except for a few occasions...

Adobe's Display Postscript was used from NEXTSTEP 0.8 to Rhapsody 5.6 (Mac OS X Server 1.2v3). Apple didn't want to pay Adobe a license fee for every copy of Mac OS X they sold. So with Mac OS X Developer Preview 1 Apple started development of a replacement originally known as Display PDF which would later evolve into Quartz.

OpenStep was simply a newer version of NeXTstep a few years later and with the ability to run on top of other OSes. Rhapsody was a version of OpenStep ported to the PPC architecture. Rhapsody/OSX Server 1.x was virtually identical to OpenStep 4.2, except I think for the display model. NeXT/OpenStep applications could be changed slightly to run on Rhapsody/OSX Server 1.x. This is why most of the early graphical OSX software was written by NeXT/OpenStep development houses.

A few notes... OpenStep is a development and application runtime environment. OPENSTEP was the name of the 4th major version of the NeXT operating system.

While structurally OPENSTEP 4.2 and Rhapsody 5.0 are very similar, Rhapsody 5.1 and later are quite different from OPENSTEP. Even applications that had been made to run on both OPENSTEP and Rhapsody act differently depending on the operating system (in my opinion they run better in Rhapsody). And some applications for Rhapsody just couldn't function correctly in OPENSTEP.

The thing I find so funny about this, is the fact that while carbon is OSX native, it's not the true native API for OSX. Apple's native API for their flagship OS was created by a company other than Apple. Though many of the people responsible for NeXT are now at Apple, Avie Tevanian for instance. Speaking of which... I understand from rather reputable sources that the Apple's purchase of NeXT was something like a reverse takeover. The story goes that the leadership roles at Apple have been replaced by old NeXT guys.

The long story goes as follows...

Steve Jobs had been relieved of any power at Apple. While still there he was unhappy with the direction of the Macintosh as it wasn't using all of the technologies that he had seen at PARC years earlier. He decided to start a new computer company to do what Apple wasn't doing with the Mac.

To start his company Jobs grabbed a bunch of people from the Mac development team. When Apple found out that all these people were leaving Apple to join Jobs they sued Jobs (he was recruiting while still working for Apple).

Apple and Jobs (NeXT) settled the suit. NeXT was not to hire any other Apple employees for a given period (I believe it was 1 year as I recall) and NexT was not to compete directly with Apple in the desktop market.

When people talk about how NeXT missed taking out Microsoft in the desktop market, they are overlooking the fact that because of Apple, NeXT could not compete in the desktop market.

As NeXT was restricted to the workstation market, and that market was evaporating due to the desktop market, NeXT was never able to fully profit from their advance technologies.

By 1995, NeXT was looking to get out of the operating system business. They had planned on passing this over to Sun who had helped in modifying the NEXTSTEP APIs into OpenStep. Sun had planned on moving from CDE to what was called Solaris OpenStep (which was an environment that felt almost identical to NEXTSTEP 3.x).

When Jobs found out that Apple was looking for another operating system on which to base the new Mac OS, he let it be known (only to Apple) that NeXT was available.

When Apple acquired NeXT, all deals with Sun were broken.

Yes, most of the main people at NeXT took over similar positions at Apple within 18 months of the merger.



So, the Cocoa APIs were originally developed by another company, which was started by former Apple employees. NeXT was never all that far away from it's Apple roots, so it wasn't surprising that it was later brought back into the fold.

 

[Steve Wozniak]

Apple/OSX will rule the computer world once it's AMD/Intel friendly

There is a reason Microsoft is $$$$$$$$. And it's not for thier hardware.

Profit margins on software are often hugely higher than hardware. Just compare the overheads. Macs today already have alot of third party parts. The perception of a slower-to-ramp architecture is hurting Apple as well. Having such a low share of the CPU market gives them little leveraging power with IBM/Motorolla.

Frankly, I wonder what apple is waiting for. I know a zillion people who would switch to OSX if they could.

NT architecture, albeit more stable than it used to be, has become a true beast of burden in usability, security, memory etc etc etc.

Cripes, if Apple marketing minds got into designing Intel PC's as well...

If I was BGates, I'd be doubling my efforts to get LHorn out the door and impressive real soon!

Game Over MS! Hello Apple OSX, the world is your oyster

ps. if only I could get my job back at apple -Woz

 

[Cooknn]

OS Nirvana

By 1995, NeXT was looking to get out of the operating system business. They had planned on passing this over to Sun who had helped in modifying the NEXTSTEP APIs into OpenStep. Sun had planned on moving from CDE to what was called Solaris OpenStep (which was an environment that felt almost identical to NEXTSTEP 3.x). When Jobs found out that Apple was looking for another operating system on which to base the new Mac OS, he let it be known (only to Apple) that NeXT was available.

Man I love that story. I fell in love with NeXTSTEP back in 1993 and again as OS X on my first Mac a little over a year ago. I dig my Power Mac G5 but it's OS X that makes Apple great IMHO Check out this 30 minute Quicktime as Steve Jobs demo's NeXTSTEP back in 1992. He disses PC's and Mac's and shows off the Dock! Q: Does anything like Database Kit exist for OS X? I laugh out loud as I watch this demo at how advanced Steve's OS was 13 years ago

 

[the Rebel]

Take a good hard look at Longhorn, windows XP SP2 patch and other Microsoft products. Anyone guess why Microsoft is taking SO long these days to come up with software?

Every change Microsoft makes has to be tested against tens of thousands of different hardware combinations. Get it wrong (and they do) and MS get bad headlines plastered all over the net. The longer Windows lasts, and the more old systems MS supports, the worse this testing nightmare will get for them and the longer it will take them to release software. Longhorn is years late - and still nowhere to be seen.

Compare this with Apple. They control the hardware, and specifically drop support for machines after a number of years. They can turn around tests really quicky, and this is reflected in the number of OS fix seeds we see, and the rapid solving of Tiger problems.

Do you really think that Apple can afford to take on the PC hardware nightmare??? Especially now that IBM's processors are finally speed-competative, and have some interesting designs for the next few years.

Yes, Apple can afford to take on the PC hardware nightmare with an x86 release of OS X. How is that? Because the x86 version of OS X would not be Apple's primary platform. The x86 OS X platform would be a bastard step child sibling to the preferred PPC platform. If there are a few hardware incompatibilities and glitches, then so what? If x86 OS X does not run perfectly on your x86 hardware, then perhaps you should buy a real Mac. Apple can afford to have the x86 version be somewhat flawed; in fact, it is to Apple's advantage to have the x86 version be somewhat less than the PPC version. There is absolutely no motivation for Apple to try to make the x86 platform rival the PPC Mac platform.