On 29/04/2021, Al Viro [off-list ref] wrote:

On Thu, Apr 29, 2021 at 02:06:12PM +0000, Mariusz Ceier wrote:

quoted

quoted

You must cause any work that you distribute or publish, that in whole or
in part contains or is derived from the Program or any part thereof, *to
be licensed as a whole* at no charge to all third parties under the
terms of this License.

The issue here is, non-GPL tools enable development and distribution
of GPL-compatible yet proprietary versions of the kernel, unless I'm
mistaken.

And?  For your argument to work, we'd need to have the kernel somehow
locked into the use of tools that would have no non-GPL equivalents
*and* would be (somehow) protected from getting such equivalents.
How could that be done, anyway?  Undocumented and rapidly changing
features of the tools?  We would get screwed by those changes ourselves.
Copyrights on interfaces?  Software patents?  Some other foulness?

I honestly wonder about the mental contortions needed to describe
something of that sort as "free", but fortunately we are nowhere
near such situation anyway.

Equivalents are not a problem - they can exist as long as the
distributed source would be buildable with GPL tools. I was thinking
that adding a requirement that the distributed kernel source should be
buildable by GPL tools would be enough to protect it from proprietary
extensions. But maybe you're right that this is unrealistic.

I don't like Rust as a language and I'm sceptical about its usefulness
in the kernel, but let's not bring "gcc is better 'cuz GPL" crusades
into that - they are irrelevant anyway, since we demonstrably *not*
locked into gcc on all architectures your hypothetical company would
care about, Rust or no Rust.

I don't mind the language. I'm more concerned about featureful rust
compiler suddenly being developed behind closed doors.