Why use Linux-Libre/GLIBC when theupstream Kernel and Clib isn't optimized to only run free Software?
The core of the GNU/Linux stack is bloated with an infrastructure that's designed around a legacy of compatibility with non-free binaries and designed around drivers that need non-free firmware and just removing non-free blobs and still using GLIBC doesn't solve the problem and it's not taking advantage of the unique position to not need that trash and makes it easier to start from scratch? Why use a compiler that has ARM support when there are no FSF approved ARM Devices? Why use GLIBC for compatibility with Debian Binaries when the distro is intended to run an unsigned whole number of non-free software that's that's measured to be less than 1?
Can we just cut the stuff that's just there to work with non-free software and maybe put back in ARM support in a Compiler designed for a Free Software only platform when there's a FSF approved ARM Board?
What's wrong with glibc?
There are less bloated and less over engineered alternatives and if you don't need to be compatible with the non-free binaries that assume GLIBC, you don't need it.
Dragora is an FSF approved libre OS which uses musl instead of glibc. You could give it a try, they released a beta for version 3.0 a couple years ago: https://sourceforge.net/projects/dragora/files/beta/
Looking through their repos, it appears to me that they keep their software up to date.
When RMS initiated the free software movement, there was no free/libre operating system. If he decided not to do anything (which necessarily involved using proprietary software) in order to maximally defend his freedom, there wouldn't be any free/libre software today at all.
Therefore such compatibility is historical legacy. You can try to remove it (and it may be as easy as you think to start from scratch, or may not). But its role in the development of free/libre software is undeniable. There has to be some compromise or sacrifice. Just accept it as an inseparable part of the struggle for software freedom.
Oh, I +1'd that!
Oh, I'm not down putting the historical legacy, I mean I think there should be software optimized for libre hardware. Maybe use something Arch based, but with a BSD kernel with all drivers for devices with non-free firmware removed and iGPUs you can't just desolder from a Skylake Core i7. I've been thinking about 80's home computers and video game consoles from a computer engineering perspective and I would say currently the newest libre implemented computer would be the SNES/Super Famicom with work on re-implementing the Sega Saturn's VDP chips (Though the BIOS would still be an issue) and even new consoles like the PS5, I think they use the BSD Kernel not just because it's non-copyleft, they use it also because they don't need to support all the stuff in the Linux Kernel. BSD adheres more to the UNIX philosophy, and sure it helps protect one from non-free extensions, but the de-facto standard is the GNU Core Utils and there's so many user made shell scripts that give people that use Mac, BSD and Busybox a hard time and there are Powershell extensions, but there was a libre implementations of Powershell before Microsoft MIT licensed it.
We do need a GPL v3 (or later) licensed kernel in the age of tivoization (Restricted Boot). However, it isn't that easy. In fact, RMS seriously underestimated the difficulties in developing a kernel. Otherwise he would consider to try to persuade Linus Torvalds to join GNU project. Once Linux kernel became a part of GNU project, it would be much easier to re-license it under GPL v3 (or later), even if Linus Torvalds betrayed GNU project like the infamous libreboot project leader did.
Current free/libre software are already well optimized for free/libre hardware. The problem is that there are so few if any free/libre hardware still being produced. Manufacturers have well known the benefits (for their sake) brought by non-free firmware, microcode, and DRM. Usually, the users don't have many options. Many of them are forced to load non-free firmware to access the Internet, for example. It is the manufacturers that should be blamed, not the users (victims) or current mainstream free/libre software and the contributors behind them.
"...even if Linus Torvalds betrayed GNU project like the infamous libreboot project leader did."
What exactly did Leah Rowe do to betray the GNU project?
I imagine nadebula.1984 refers to Leah Rowe's decision to exit GNU little after she made Libreboot a GNU package: https://lists.gnu.org/archive/html/info-gnu/2017-01/msg00001.html
What happened should not matter anymore, in my humble opinion. Here is Leah's apology: https://libreboot.org/news/unity.html
Three weeks later, the Libreboot was re-submitting an application to re-join GNU: https://libredd.it/r/libreboot/comments/66tdds/proposal_for_libreboot_rejoin_gnu_community/
"optimized to only run free Software"
This is an impossibility to make a computing environment that can "only" run free software. Computers are general purpose devices that will run whatever code given to them, while the question of whether a program is free or not isn't a technical issue that a computer can decide. So make up any computing environment you want. I'll be able to write a non-free program for it.