On Tue, Feb 4, 2025 at 1:46 PM Danilo Krummrich <dakr@xxxxxxxxxx> wrote:
>
> On Mon, Feb 03, 2025 at 03:24:10PM -0500, Joel Fernandes wrote:
> > Hi Danilo,
> >
> > On Fri, Jan 31, 2025 at 11:04:24PM +0100, Danilo Krummrich wrote:
> > > +#[pin_data]
> > > +pub(crate) struct NovaCore {
> > > + #[pin]
> > > + pub(crate) gpu: Gpu,
> > > +}
> >
> > I am curious what is the need for pinning here in the patch in its current
> > form, is it for future-proofing?
>
> Yes, it is.
>
> It's not always needed, but since I know that further down the road we'll need
> at least a few mutexes, it seemed reasonable to already consider it.
It seems reasonable to me as well, I would probably also add a code
comment there about what is expected to be unmovable in the future
(Just in case any code readers don't raise the same question I was
raising).
The source of the confusion for a reader could be the documentation
expecting a #[pin].
- Joel
