Well, now that the dust is beginning to settle and the celebratory drunken debauchery is merely a happy memory (or no memory at all, if you did it right), one thing has become abundantly clear: while Panther is, in fact, all that and a bag of chips, said bag of chips is not necessarily, say, a jumbo-size bag of Baked Lays. In at least some circumstances, the bag more closely resembles a snack-size vending machine dispensation of soggy Baken-Ets. To put it in a far blander and less snack-related manner, everybody loves it like crazy, but Panther's got problems, and we can hear the wheels turning for a much-needed 10.3.1 update already.

Yesterday we mentioned that less than a day after the installation of Panther on our almost embarrassingly vanilla original 12-inch PowerBook (no third-party RAM, no system hacks, no shareware, and only a few commercial applications installed), the poor thing suffered its first kernel panic ever. Well, now the count is up to two panics in a three-day period, following eight straight months of panicless Jaguaring. Both occurred as we tried to wake the 'Book from sleep, implying that we're looking at a bug, not a fluke. And while it's true that the Panther kernel panic is every bit as sedate as the Jaguar version, a forced reboot and the loss of all unsaved changes is likely to be annoying no matter what it looks like. Indeed, being calmly informed that "you need to restart your computer" is almost more infuriating than the classic vomitous spate of ASCII technojargon and hexadecimal code that used to come spewing through the GUI yelling "OH MY GOD!! OH MY GOD!!" (After all, if you're going to panic, then panic already.)

Well, a quick glance around the 'net shows that we're not the only ones having "issues" with Panther. MacFixIt is rife with reports of fun little problems like data on FireWire hard drives getting scrambled beyond redemption, FileVault nuking preference settings, and Exposé appearing to play fetch with the family dog but in actuality only pretending to throw the ball and thus confusing the dickens out of poor Fido. Meanwhile, WIRED's Leander Kahney reports having spent the entire weekend on data recovery tactics after the installation of Panther on an iBook went so far south it saw penguins. (Despite the pain, Leander still gushes over how badly everyone needs this upgrade.)

Yikes! Suppose this is what happens when you release software a couple of months before the deadline? Looking back, we're actually pretty darn happy that we plunged headlong into that evil 10.2.8 update without testing the waters first, because the resulting ickiness from that process made us just nervous enough to be able to avoid the nigh-irresistible temptation to Pantherize our main production Macs the very second that black box with the shiny X showed up on the doorstep last Friday. So for now, for the mission-critical stuff (like AtAT production, nuclear facility failsafe design, and Elmo's Keyboard-o-rama) we stick with Jaguar. And watch. And wait.

Meanwhile, there's another little downside to Panther's release: the free public beta of iChat AV is no longer available from Apple's web site. iChat AV is now officially at 1.0-- and as promised, it's free with Panther, but Jaguar owners who want it without upgrading their systems to Apple's latest big cat are going to have to shell out $29.95 for the software. (If you're already running the beta, it won't expire until the end of this year, so you've still got some mileage left on the free ride.)

As far as we're concerned, that's all well and good, since Apple laid all this out when it introduced iChat AV last June. But there is something about the situation that sticks in our craw just a bit: contrary to all sense and logic, iChat AV is not included with the iSight.

No, seriously. It's not.

Really.

We swear, we're not joking.

Quit smiling like that! Sigh, okay, here-- check for yourself. According to Apple's web page, iSight is "Made for iChat AV," it "Requires iChat AV," but iChat AV is "not included" with the iSight. See?

Pretty [BLEEP]ed up, huh?

We're not the only ones who think this is goofy, are we? The upshot is, every single iSight ever sold prior to Panther's release (probably a lot-- we remember it was topping the Apple Store Top Ten for quite a while) is going to wind up useless come January 1st unless its owner shells out more cash on top of its original $149 purchase price: either thirty clams for iChat AV 1.0 or a hundred more than that for the full-on Panther upgrade. And anyone with Jaguar right now who runs out and buys an iSight at an Apple retail store is going to come home to discover that he or she just spent a hundred and fifty smackers on a piece of hardware that serves only as a snazzy-looking paperweight until still more cash gets spent.

Call us pathetically naïve ("You guys are PATHETICALLY NAÏVE!"), but we really expected at the very least that new iSights would come with a copy of the software that actually makes it work. Heck, we're even crazy enough to have thought that all registered iSight owners would be eligible to download the software for free-- call the men in white coats. Instead, people who went out on a limb and supported Apple by purchasing a new product that does just one thing are now finding out that it won't actually do that one thing unless they also buy the software that drives it. We don't sleep much, so it's possible we're missing something here, but doesn't this strike anyone as just the slightest bit... well... sleazy, ungrateful, and insulting to the customer base? Just wondering.

And the numbers just keep on climbing! Yes, kiddies, it's time for the almost-daily Big Mac Report, which tracks Virginia Tech's G5-based supercomputer as it slowly climbs the greased pole of its own limitless potential. Well, okay, maybe not limitless, exactly-- its theoretical max performance is 17.6 teraflops, and exceeding that would require cold fusion, a means of exceeding the speed of light, and the blood sacrifice of a live virgin to the ancient and terrible volcano god Kajagoogoo. But hey, never say never.

If you can still remember anything from Friday before the moral turpitude and massive abuse of alcoholic beverages at Night of the Panther, you might recall that, after first turning in a relatively pathetic score of just 7.41 teraflops earlier in the week and then improving that to 8.164 teraflops just a day or two later, Big Mac finished out the week with a darn respectable performance of 8.7 teraflops. That just barely brushed past the best-scoring Intel-based supercomputer to make Big Mac the third fastest supercomputer on the planet-- provided that nothing changes before the official scores are recorded next month, that is. The Intel cluster scored 8.633 teraflops, so the battle for third place is a horserace.

Or rather, it was. Faithful viewer Dan Birchall informs us that things are changing-- for the better. Jack Dongarra's report has been updated yet again; now it's stamped with today's date, and if you download the PDF and flip to page 53, you'll find that since last Friday, Big Mac has gone from 8.7 teraflops all the way up to a stunning 9.555 teraflops. If the Intel cluster has improved at all in the meantime, it's not reflected in Jack's report, so as of right now, it looks like Big Mac has carved out a nice, cozy spot at Number Three.

Unless, of course, Big Mac somehow manages to claw its way into second place. It's a tough climb, to be sure; that spot is currently occupied by an ASCI Q AlphaServer EV-68 (there are three of 'em in the top ten) with 8,160 processors-- nearly four times as many as Big Mac has. The ASCI Q has a recorded score of 13.88 teraflops, which is a pretty lofty target for Big Mac to hit. Can Virginia Tech squeeze four more teraflops out of Big Mac? Like we said, never say never; after all, they've already dragged out two extra teraflops in just the past week. Also, four percent of its processors are still apparently offline (it's listed as a 2,112-processor cluster, so there's another 44 Power Macs still warming up in the bullpen or something), and we've heard from a few alleged Hokies who claim to be in the know: reportedly there might still be one or two obvious bottlenecks to be ironed out in the coming weeks' tweaking.

It's not terribly likely, mind you, but if Virginia Tech can somehow get Big Mac crunching at 80% of its theoretical peak (the Earth Simulator does 87.5%-- of course, that's not a cluster), then, yes-- there's a slim chance of slipping into second place. Dare to dream!