Comment 21 for bug 1847105

Yep Daniel, just found the same discussion a few minutes ago and updated the Ubuntu bug.

While correct, your suggestion of checking pool features for this seems very complex for the trivial issue that we see atm.

Maybe (only maybe) virt-manager would want to sparse-alloc always and could then change to not request allocation==capacity.
From most common virt-manager use cases I know of sparse would make more sense and be faster (at creation).
I'd expect non-sparse to be a special case that maybe gets a way to be specified if really needed.

And also on the libvirt side we still might consider reverting the commit that broke behavior which would render all the effort to flag those features at pools not needed (at least not for this).