Friday, August 23, 2002

eWeek eats own foot

(via MacRumors)

eWeek is reporting that:

They said Apple is planning to "block" next-generation Macs preloaded with Panther from booting into Mac OS 9 although the company will continue to support applications running within Mac OS X's Classic environment.

Talk about uninformed or just plain stupid reporting -- they make it sound like Apple is going to expend effort to make OS 9 stop booting natively.

First, whether or not a machine can boot Mac OS 9 has nothing to do with whether or not it has some random version of OS X installed. Mac OS 9 boots independently of OS 9. Certainly, Apple could make the OS X installation such that it would effectively not boot if OS 9 had been previously installed or vice-versa, but a low level format and resetting the nv-ram will destroy all evidence of a prior operating system.

But, the real stupidity is this: Does eWeek have any clue as to how much it costs to port OS 9 or OS X to a new architecture? Answer: A lot -- it requires a tremendous engineering and testing effort.

I would be willing to bet that OS 9 is significantly more expensive to port both because it has less seperation between applications and hardware and because of the incredibly vast quantity of legacy code within the codebase. Because OS X is a sharply layered system, successfully porting the lowest layers quickly leads to a working system. For OS X, it is the same situation as Linux -- how many times have we seen some random person announce "I booted the Linux kernel on my DreamCast / Palm / Microwave" only to see full distributions shipping within weeks?

Even if OS 9 costs the same or is even easier to move forward to new hardware, what about all of the cost to continue maintaining the legacy codebase? Or the cost of continuing to perpetuate an operating system that requires two codebases for every iApp and anything else Apple wants to ship on it?

(Which brings up another sore point: people don't realize how much more it costs to build apps for OS 9 than OS X. It's the APIs, stupid! The OS X APIs -- Cocoa and the Core, in particular -- are leaps and bounds beyond any other app development environment on the planet. Sherlock 3 -- a Cocoa app -- comes naturally in OS X, but would be a nightmare to build on OS 9.)

So -- there is a very real, and very expensive, technical issue that leads to the question: How long does Apple keep investing in porting OS 9 to new hardware?

Simple answer: For as long as the return on investment justifies such effort.

How much revenue is generated by machines sold solely as OS 9 workstations? Certainly, the music community is very much in the situation that the required software/hardware has, for the most part, not been ported to OS X. But that is starting to change -- that market is decreasing in size.

Furthermore, the design of the hardware is such that OS 9 can't take advantage of it anyway. For most OS 9 users most of the time, a dual processor machine is useless -- OS 9 simply doesn't take advantage of multiple CPUs at all. This is exactly the opposite of OS X -- it doesn't matter if an app is specifically written to take advantage of multiple CPUs (easy: add a thread for calculation, for example), OS X will still take advantage of that additional CPU horsepower.

Going forward, there are architectual changes that Apple could make that would require major changes to the internals to OS 9.

At some point in time, it doesn't make business sense to move OS 9 forward to new hardware simply because it doesn't make technical sense.

For Apple's sake, I hope that time is rapidly approaching -- OS X provides the flexibility for Apple to make some radical changes on the hardware front. I hope such an opportunity has presented itself and Apple is pursuing it with alacrity. That OS 9 will no longer slow them down is a huge advantage.

(Does anyone actually read all of this or am I just flinging keystrokes into the empty ether?)
9:21:05 AM