If It Ain’t Broke

Friday, 23 April 2004

Around a month after Mac OS X 10.3 shipped, I charted its adoption rate among Safari-using Daring Fireball readers. (This tracking is possible because the Panther versions of Safari aren’t available for Jaguar.) The gist is that about half of you upgraded within the first few days of Panther’s release, and within a month, somewhere between 75 to 80 percent of you had upgraded.

The stats for the first three weeks of this month put the current number at just under 90 percent (89.675 percent, for you stat nerds).

From its debut, Panther has been almost universally lauded as a significant and worthy improvement over Jaguar. So, who are these people who have upgraded as of April 2004, but waited several months before doing so?

Some of these people purchased new Macs. If you were planning on a new machine, why bother paying for a copy of Panther to upgrade the machine you were getting ready to replace?

Most of these Panther-come-latelies, however, are just plain cautious. People with work and deadlines, and for whom a stable and reliable Mac running an older version of the OS is just fine, thanks — deadlines don’t get moved just because your OS upgrade went awry.

Exhibit A: Jeffrey Zeldman, who upgraded from Jaguar to Panther earlier this month. It did not go well:

My journey into Panther killed my Titanium PowerBook in stages. First came software failure: Apple applications such as Safari quit on launch; the machine could not find the network. Then came kernel panics. (This is where the machine reboots into a black and white Unix screen, spitting out Matrix-like error messages. To exit, you must type the appropriate Unix commands, which implies that you know what they are.) Finally, the machine would not boot, period.

It took three days’ heavy digging to restore the Mac to operability. Along the way I discovered a couple of Apple design decisions that make sense under normal circumstances but turn problems into disasters when things go wrong.

His report detailing the entire experience is exquisitely detailed, and well worth reading — even if you, just like me and the vast majority of Panther users, upgraded to Panther without a hitch.

The fact the Panther upgrade worked perfectly for you, and for most of us, is not proof that Mr. Zeldman must have done something wrong.

Nor is the fact that something went wrong for Mr. Zeldman an indication that Apple “doesn’t test” their updaters, or that they have rampant QA problems.

Bugs happen. Some will slip through even the tightest QA tests. It has always been the case, and always will be, that every upgrade of your OS ought to be preceded by a full backup.

Adding insult to injury, after he finally got Panther installed and running, Zeldman allowed Software Update to bump the machine to 10.3.3, and he ended up unable to print. The problem is that for upgrades, the Panther installer doesn’t create several new users, necessary for Panther, but which weren’t needed in previous versions of Mac OS X. Running Disk Utility’s Repair Permissions command can then leave you unable to print, because files which are supposed to belong to these new “users” can’t be assigned to them because the users don’t actually exist.

I ran into this problem myself shortly after upgrading, and All OS X’s Panther Printing Fix utility corrected the problem. Unfortunately for Mr. Zeldman, this utility only restored his ability to print to a networked printer; he’s still unable to print to a USB printer connected to his machine. (I would have suggested this solution to Zeldman, but I’d forgotten about my similar experience until he mentioned it in his aforementioned post earlier today; I no longer remember trouble-shooting tips, I trust Google to remember them for me.)

At the end of his first article detailing his Panther saga, Zeldman writes:

As a consumer platform, OS X is years ahead of the competition. As a platform for computer professionals with a solid Unix background, OS X is also years ahead. But I wonder if Apple has lost sight of the non-Unix-oriented creative professionals whose loyalty supported the company through its hardest times. There are many of us. We admire what Apple designs, we remain committed to the platform, and we want the company to succeed. But a simple OS upgrade should not fail, should not induce panic, and should not waste three days of a user’s life.

It’s also the case that design professionals tend to purchase Apple’s higher-margin, top-of-the-line hardware. They’re not just loyal Apple customers, they’re profitable Apple customers.

More to the point, however, is the uncomfortable fact that many print design professionals are still using Mac OS 9 (or even 8.6). There is a huge “it just works” advantage that Mac OS 9 holds over OS X, and Zeldman’s recent experience exemplifies it.

OS updates didn’t always go perfectly on the old Mac OS, but when they went wrong, they very seldom left your Mac unbootable. And small point updates — going from X.x to X.y — tended not to mysteriously break your ability to print.

When things went wrong on Mac OS 9, you could suss out the problem by dragging files into or out of the Extensions folder and restarting.

When things go wrong on Mac OS X, they often happen at a deeper level. File permission and ownership problems, for example, are not something a typical Mac user can deal with. It’s not enough that Mac OS X doesn’t require any Unix nerdery whatsoever for day-to-day use — it should never require Unix nerdery to recover from software updates, either.