On Tue, 13 Sep 2016 16:33:25 +0100 Chris Lamb <la...@debian.org> wrote:
> Hi Reiner, > > > If I remember correctly, the reason why reversing the order is the > > default, was that this guarantees a different order. Reiner's recollection is correct. > Whilst this is true and really useful feature of disorderfs, it's > more that something called disorderfs isn't actually introducing > disorder for any reasonable definition - ie. it's reversing. That's true, but on the other hand, if randomization results in the same order (which is a very real risk for small directories), then disorderfs is failing at its description in an even worse way, by not changing the dirents order at all. Perhaps we could guarantee that the shuffle results in a different order? I think that would resolve any objection to making shuffling the default. Regards, Andrew
