Quoting Craig Sanders (cas@taz.net.au):

i.e. label collisions are not an accident, they're the inevitable result of poor planning.

Sure.

The trouble is that unless you want to use UUIDs there isn't any alternative.

Well, speak for yourself. I've been using /dev/sdX (never used IDE ;-> ) on Linux since 1992, and for whatever reason never encounter the variety of ills you cite.

There's not much use to a "unique" identifier that is only "unique" some of the time, or only unique as long as any humans involved don't do the kinds of things that us humans are prone to doing.

Personally, I'd settle (or consider settling, depending on outcomes) for a human-compatible identifier that's roughly as unique as an SHA1 hash is, or maybe even an order of magnitude or so less so. As I'm sure you know, those are not guaranteed absolutely reliable as a digest form of the things they stand for, but for all _practical_ purposes they are. For purposes of my disk identifiers, it really doesn't bother me that a few thousand other disks somewhere else on the planet might at some given time have the same identifier. As the late Adam Osbourne or Osbourne Computer used to say, 'Adequacy is sufficient.' (Hey, I'm old, and I was a member of the Homebrew Computer Club as a teenager.)

So it makes sense to give a UUID to disks/partitions/filesystems by default.

You keep telling yourself that. ;->

and then one day you'll compile the latest shiniest new kernel and discover that the kernel devs really weren't joking when they said you can't rely on the names to remain the same.

You keep saying that. It's been 26 years.

I see it frequently. But my main file server has 8 SATA drives plugged into an LSI SAS controller and a few more in SATA ports on the motherboard.

Yeah, personally I'd carefully avoid that.

mdev already has helper scripts that create the same kind of /dev/disk/by-* symlinks that udev does.

Well, I'll not use those. ;->