On Sat, Apr 13, 2024 at 03:28:31AM +0800, Zorro Lang wrote:
> > I wonder, how many tests should keep going if _scratch_mkfs_sized isn't
> > successful? Would it be simpler to make a _try_scratch_mkfs_sized for
> > those callers, and make the main _scratch_mkfs_sized simply do
> >
> > _scratch_mkfs_sized()
> > {
> > _try_scratch_mkfs_sized $* || _fail "_scratch_mkfs_sized failed"
> > }
> >
> > ?
>
> I agree with this. Actually I don't know what's kind of historical reasons
> make _scratch_mount _fail directly if mount fails, but _scratch_mkfs
> doesn't like that?
Agreed. Unsupported mkfs should absolutely fail by default. For some
tests instead of _fail it should be _notsupported, but that would be a
separate helper and between the two of them we'd cover almost all uses.
