Paul left a couple of thoughtful comments about Leopard: Hasn’t Apple been swatting the dual-platform bugs all along? Won’t Time Machine eat my hard drive?

Apple has, indeed, been, secretly and openly working on the Intel code almost since Apple acquired NeXT. However, Apple has a finite OS engineering team (there were about 300 OS engineers during my tenure there in the mid-late 90s). Like most firms with a popular product that has to be delivered on 2 platforms, they have devised a development method that abstracts as much of the product as is humanly possible to code that ‘juts runs’ in both environments.

Apple, unlike most application developers, can choose to make their home-brew dual-platform solution the development environment that everybody else has to use. They call it Xcode. I’m in a position where I can appreciate how painful it can be to leave well-debugged, and well-understood, development tools for new ones. This must have been as hard at Apple as it was anywhere else, and they seemed to have pulled through (though one wonders if the user experience with OS X 10.4 – it was very buggy at launch – didn’t reflect engineering productivity lost to the switch).

The hardest part of creating useful commercial software is not usually writing the software: the hard part is testing and debugging it. Companies that hope to be profitable, have to put some limits around testing… if you make a matrix that represents every possible combination of configurations that a modern user could possibly have, you rapidly descend into combinatorial neverland, where your testing schedule would outlast the universe by some very large amount of time. When did you want that next rev? 20,006 good for you? I wonder what kind of hardware we’ll have 18,000 years from now.

So, you have to put some limits on testing, lest the widows and orphans who depend on your stock’s performance suffer. One thing every cross-platform developer knows is that some bugs are the same, and have the same fix, on both platforms. That’s a good thing. Some bugs, however, are different, and have different fixes, sometimes very, very different fixes, on each platform. But your engineering and QE team is finite: you have just so many guys and gals and hours in a day. So, you have to make choices. Do PPC users really need this? Intel users?

The point I’m trying to make is that Apple has probably more experience than anyone else on PPC/Intel cross-development, at least for Mac OS X. But they still have to make choicess, which divies up at least part of their engineering and quality engineering teams into platform-specific specialists, which slows top-level development.

On the Time Machine issue, Paul raises a good point, given my experience with Apple’s Backup product. It’s true that locally (meaning anywhere the Internet is available), 500 GB drives can be had for $190, which suggests that this will only be a problem for feature-length video producers and maybe some genome hackers. Prolific still photographers and parapatetic writers and copy editors, you’d think, wouldn’t be seeing problems over the life of their machine.

But, to Paul’s point, I’ve been rolling my Mac home directory forward since Mac Os 5.x days. It has ballooned tremendously since I started using digital cameras and iTunes: tonight it uses 47 GB, including a 20,000-picture iPhoto archive, a 15,000-picture iView Media Pro archive (including scans of pix I took, as an amateur and professional as far back as the 1960s) and gods-know-how-many iTunes files. My (Apple) Backup incremental archives tend to be at least 1 GB, often bigger. Nightly backups chew up around 50 GB a month, which likely fllls my cheap ‘giant’ HD in the first year or much sooner. Let’s hope Time Machine is better, and/or more selective than Backup (or just plug a bunch o’ drives into your Mac Pro). Alternately, buy a cheap Terabyte drive next year…

{ 2 comments }