Nouveau - Nov 2025 - [PATCH RFC 1/4] rust: clist: Add abstraction for iterating over C linked lists

On Tue, Nov 4, 2025 at 2:42?PM Alexandre Courbot <acourbot at nvidia.com>
wrote:
>
> What I'm more worried about is that it might be a PITA to write. :/ But
> maybe the core folks have an example for how this could be done cleanly
> and in a reusable way (i.e. we don't want to duplicate the dummy list
> creation code for every example).
Using a shared module/file may be good enough, as long as the `#[path
= ...] mod ...;` or `include!(...)` is hidden with `#`, i.e. as long
as the user does not need to see that to understand the example.

But, yeah, we have already a few places in the tree with fake `mod
bindings` for doctests and things like that.

Cc'ing Guillaume in case there is a better way to do this. The "have
something applied to several parts of docs" has come up before for
references too (the "external references map" I proposed).

In any case, even if the example does not run, it is still way better
to have it at least build instead of completely ignored, because that
means it will not become stale.

Cheers,
Miguel

You can use `cfg(doc)` and `cfg(doctest)` to only include parts of the
docs when running doctests (if that's what this is about).

Le mar. 4 nov. 2025 ? 15:07, Miguel Ojeda
<miguel.ojeda.sandonis at gmail.com> a ?crit :
>
> On Tue, Nov 4, 2025 at 2:42?PM Alexandre Courbot <acourbot at
nvidia.com> wrote:
> >
> > What I'm more worried about is that it might be a PITA to write.
:/ But
> > maybe the core folks have an example for how this could be done
cleanly
> > and in a reusable way (i.e. we don't want to duplicate the dummy
list
> > creation code for every example).
>
> Using a shared module/file may be good enough, as long as the `#[path
> = ...] mod ...;` or `include!(...)` is hidden with `#`, i.e. as long
> as the user does not need to see that to understand the example.
>
> But, yeah, we have already a few places in the tree with fake `mod
> bindings` for doctests and things like that.
>
> Cc'ing Guillaume in case there is a better way to do this. The
"have
> something applied to several parts of docs" has come up before for
> references too (the "external references map" I proposed).
>
> In any case, even if the example does not run, it is still way better
> to have it at least build instead of completely ignored, because that
> means it will not become stale.
>
> Cheers,
> Miguel