(Despite the fact that I ran Linux for years on a laptop where suspend/resume didn’t work, now that I’ve had it working for a few years, the prospect of living without it again is just too painful.)

That comprises its strength; but it’s also a huge weakness. Minimal redundancy makes metadata corruption hard to detect and likely to be spectacularly disastrous. The filesystem is designed to be compact and very fast, but it’s brittle and incurs a heavy CPU tax for the speed you get.

That’s the opposite set of tradeoffs than I’d generally prefer in a filesystem: I’d like it resilient and easy on the CPU. The perfect choice would be the nigh unbreakable ext2, of course – if only fsck didn’t take forever and 3 days to finish on modern harddisks. So I switched to ext3 partly out of laziness, and partly because at the time, none of the other alternatives to ReiserFS were any good. I’ve heard quite a few good words about XFS and some about JFS bt now, but I’m not sure how well they’re shaken out so I’m languishing on the safe default. As an aside, I wish there was a good modern filesystem that can be read and written by both Linux and FreeBSD…