On Tue, Oct 17, 2023 at 06:25:07AM +0200, Christoph Hellwig wrote:
> On Mon, Oct 16, 2023 at 11:41:24AM -0700, Darrick J. Wong wrote:
> > > +_require_scratch
> > > _require_scratch_delalloc
> > > _require_scratch_reflink
> >
> > Oops. That _require_scratch_delalloc could've gone beneath the
> > _require_scratch_reflink to take advantage of the _require_scratch call
> > in _require_scratch_reflink.
> >
> > That said ... maybe _require_scratch_* calls should require the caller
> > to have _require_scratch'd beforehand? And perhaps all
> > _require_scratch_* functions should complain if
> > ${RESULT_DIR}/require_scratch doesn't already exist.
>
> Either having a _require_scratch in all _require_scratch_* helpers or
> none sounds sensible to me, having it in some and nother others is
> rather confusing. Any preferences?
IMIO, every test should be calling _require_scratch explicitly, and only
once. The _require_scratch_XXX functions should complain if you haven't
called _require_scratch.
--D
