On Sat 22 Apr 2017 07:56:57 PM CEST, Max Reitz wrote: >> So, if you got this far in reading, the question becomes whether >> having a mode where you can mark a cluster as >> mapping-reserved-but-unallocated has enough use case to be worth >> pursuing, knowing that it will burn an incompatible feature bit; or >> if it should be rolled into the subcluster proposal, or whether it's >> not a feature that anyone needs after all. > > I just forgot that just saying !ALLOCATED will be enough, regardless > of the ZERO flag... Yeah, subclusters will give us this for free, and > I think it's therefore reasonable to just require them if you want > this feature (preallocation with a backing file).
I agree with Max here. Berto