The falling myth of Apple reliability
Some latest Apple's Mac OS X updates have caused severe troubles to users because of introduced instability. One of them is related to the capability of running some Java applications.
Let me just sum up what happened to me in the latest days. I was in Stockholm for a business trip and had the insane idea of violating one of my best practices: "You won't update your operating system if you don't have your lab at hand". And in order to keep my luggage light, I also violated another: "You won't ever leave without taking with you at least two items from: o.s. CD installation set, external hard disk with a clean bootable o.s., backup iBook 12". Forgive me, father, for I have sinned; I shall never do that again.
Basically, after the update to
Mac OS X 10.4.10 QuickTime 7.2, my MacBook Pro refused to reboot. I wasn't completely blocked in my job only because I had Linux in the multi-boot. What I found out is that some files, including a couple of very-basic libraries, were zeroed out thanks to this bug. As you can see it's a very old bug, and BTW it's not Java related. In facts, this was only the first half of my adventure and it's not the main topic of this post.
After finding and fixing the corrupted files (not easily: I found out that installing the update on the clean Mac OS X running on the external drive screwed it too!), I found myself with some Java applications not running, most notably FreeMind and MagicDraw.
Basically their icon just bounced once in the dock and then died. After some tries I was able to track this down to a strange error message in the system logs and to relate it to the unability of launching Photoshop Elements 3: indeed I guessed the problem was with Rosetta, the PPC emulator which comes with Intel Macs.
Indeed Rosetta appears to be related to the execution of some (old) Java applications. Mac OS X deals with Java Applications packaged for it by "wrapping" them with a special launcher, named JavaApplicationStub, which in turns invokes the JVM. Recent Java applications have Universal Binary stubs (that is, they contain both PPC and Intel code), while older ones only have PPC code. In this case Rosetta is started, even though the native VM is then invoked. But if Rosetta is broken, you can't run those Java applications with a PPC-only stub.
While it took me a couples of days to fix this, only a few minutes ago I found that MacFixIt has just posted an updated report about this bug. Basically the problem comes with QuickTime 7.2; the solution is to first revert to QuickTime 7.1.6 (by using a tool cited in the report), then re-install the latest Mac OS X update. BTW, there are reports saying that the bug is triggered by some extra conditions such as having the Apple Java 6 pre-release installed, but - believe me - I didn't have it.
Now everything seems to be ok again, but for some native libraries in the Java installation that have been probably zeroed out by the prebinding bug. I'll copy them tomorrow from a working installation of some co-worker.
Now, I can't prevent myself from seriously complaining about Apple. I note that:
- A Software Update has been clearly rolled out with serious bugs inside (breaking Rosetta for many users means to break Microsoft Office 2004, Photoshop stuff and many other applications - something that can't go unnoticed in serious testing). Put this together with the fact that Apple has been forced to withdraw another buggy update about the optical unit, while a third update is giving troubles and you can easily infer that Apple has moved too many engineers from the Mac OS X department to the damn'd iPhone.
- In spite of the Rosetta problem being around since several days now, AFAIK Apple didn't post any technical advice so far, neither announced a further fix to address the problem.
This sounds too Microsoft-like - and I've said it all. Add to this that I've been forced to buy a new power supply for my MacBook Pro since the cable melt down due to a faulty insulator, and that the new unit after four weeks is starting breaking again in the same point; and that I'm hearing more and more friends and co-workers complaining about small or medium hardware troubles with their unit. Well, since I didn't buy Apple for the griffe, but because I wanted a system that didn't make me waste time or money, you understand why I am seriously disappointed.
This is the last serious Apple trouble that I'm forgiving. At the next one I'll start planning migrating to Linux Ubuntu.