Working myself in the embedded industry, I know that the vast majority of embedded systems (including ones with a single dedicated CPU) run ancient versions of the kernel. In the case of such systems being finally updated to the latest kernel their code would need changes anyway (and in most cases the whole code is rewritten with modern technologies). More than that, the new embedded systems, even the cheapest ones, have at least dual core CPUs.

Thus, kernel leadership should not be afraid to deprecate and remove obsolete things. This is the purpose of deprecation, namely updating of the old code to the new reality. Just adding guards in the man page is the lazy and unsafe approach which provides work for the Linux kernel consultants.

Most issues including serious security issues are caused by people being afraid to remove unsafe / obsolete functions because it could break old code. But modifying those functions breaks the code in a hidden way, which is far worse. This is why the standard C library still has unsafe string functions (and many others) which are exploited by hackers every day.

We very often defer the task of providing warnings to the compiler, instead of forcing the API and language creators to do their job.

This is why, I consider that such a tone used by Linus Torvalds who is the leader of all kernel development, is inappropriate, because such historical garbage has accumulated under his leadership, thus, he bears at least part of the blame for the wrong usage of such "historical garbage" which was left in plain sight for the last 10 years or so.

And this is why, two persons spent some minutes today, which could be used doing something more useful. Thank you for a really nice conversation!

Alex