Btrfs: broken file system design (was Unbound(?) internal fragmentation in Btrfs)

Or are usually slow but can be incredibly fast :-) Hence the need to know your data!

The sort you are thinking of is the quick sort - under most circumstances it's the fastest.

The one I'm thinking of is the bubble sort :-) The watermark-optimised version, run over a already-sorted input set, is provably the fastest sort possible! And this is also the quick-sort worst case!

I'm a regular user of the bubble sort and variants, but oftentimes I know my datasets are approximately sorted before I start.

Cheers,
Wol