> Seriously, you asked an excellent question. The loop driver will
> end up causing "block not present" errors when accessing a file
> with holes [I just tried it]. At least it's not fatal.
>
> I don't know if this is correct behaviour, or fixable.
> I'm open for suggestions.
IMO: Allocate a block if there's space, something like the current 'badspot'
behaviour if the block can't be allocated (eg 'cause of a badspot ...)
While you're on the subject of new features, is there any hope of allowing
a loopback device onto /dev/mem (Yes the _character_ the device) or an
equlivent ?
It was one of the most obvious things to try for >1.2.13 when I wanted
to put a swap file onto the slow memory in an old 386...
-- Rob. (Robert de Bath <http://www.cix.co.uk/~mayday>)