On Thu, 3 Feb 2022 at 21:50, Miguel Ojeda via Liaison
<liaison_at_[hidden]> wrote:
> Of course there is a chance -- as with any other proposal. The
> no-identifier form does have pitfalls depending on how a compiler
> defines it, but it is widely used everywhere.

That's a rather bold claim to make, considering that there are
union/bind mount scenarios where
it doesn't work, or requires implementations to do full-content
hashing to be able to say that two
apparently different files have the same content. Such scenarios were
explained with sufficient
plausibility to WG21 in 2012 that a suggestion to standardize #once by
that committee was dropped.
Nothing in the current WG14 proposals even remotely suggests that
solutions have been found
to such problems.

So no - it's not widely used everywhere, there are quite a few
software houses where its use is completely banned,
because it doesn't work.