On Mon, 1 Sep 2025 17:16:56 +0100
ael <witwall3@???> wrote:
> On Mon, Sep 01, 2025 at 11:05:53AM +0200, Davide Biondi via Dng wrote:
> >
> >
> > my name is D. I am not a computer scientist, but I have been a
> > long-time Linux user with some entry-level coding experience.
> >
> > 2-The Rust toolchain is complex, not fully bootstrappable from 100%
> > free sources, and relies on binary distributions. This reduces
> > auditability and increases dependency on external actors.
> >
> > 3-The expansion of Rust in the kernel and in base utilities may bind
> > the Linux ecosystem to non-fully-free toolchains and reduce long-term
> > freedom for users and developers.
> >
> > 4-Combined with the ever-growing scope of systemd, this convergence of
> > design and licensing risks turning Linux into something more
> > monolithic and Windows-like, moving it away from the traditional
> > Unix philosophy.

I share your concerns, especially because as Linux moves from a user
developed and maintained set of code-bases to a corporate developed and
maintained set of code-bases. Rust, as you may recall, is a Mozilla
foundation.

I have some ideas for how to solve this problem, but they'll probably
have to wait until we solve the problem of corporate controlled firmware
that is totally insecure (in the sense that it's never been audited),
is able to access the internet without permission, and, in some cases,
doesn't allow Linux to boot. Although Intel's IME counts, I'm more
concerned about Microsloth's Pluton.

> While I share some of our concerns, I have to say that if you had more
> knowledge of computer science, you could not do other than welcome Rust
> as a huge advance on C. It is not ideal, but there is a reason that it
> has been accepted into the kernel.

I've read this sort of thing over and over again ad infinitum. Why not
use Python, Ruby, Pascal, SH, etc., in the Linux kernel if you want memory
safety?

No one ever did a review of the various languages available, even rust's
author, and demonstrated that rust is, overall, the best replacement for
C out there. Did you not know that Pascal has been the #1 recommended
replacement for C/C++ code for years before rust showed up? There's a
whole C string library out there licensed CC0 and no project ever used
it! We've had options to help with C code guys!

In fact, Linus Torvalds himself said he did not want object oriented C++
code in the Linux Kernel. Now we're doing OO with rust! Why not let C++ in
then!? We have the STL, so no need to complain about C++ oddities of the
past, and we have libgc, so there's little need to complain about memory
leaks either.

The reason is all too simple, corporations, including Mozilla foundation,
wanted rust code in stuff, so they're pushing for it, and, for whatever
reason, Linus has surrendered in this and many other social/corporate
matters.

IDK what these big corporations intend, but this whole thing reeks of the
old EEE (Embrace, Extend, Extinguish,) campaign. Once they control the
major code bases and have displaced most of the freelance devs, we'll be
at their mercy. A great way to do this is to increase code complexity, by
using tighter SW coupling, systemd, and using more programming languages
in projects, rust (and thus requiring rewriting the old, known good, code
in the process).

David