Hands have to get dirty somewhere. "As deep as The Worker's City lay underground, so high above towered the City of Metropolis."
The choices are:
1. Restrict the freedom of CPU designers to some approximation of the PDP11. No funky DSP chips. No crazy vector processors.
2. Restrict the freedom of OS designers to some approximation of Unix. No bespoke realtime OSes. No research OSes.
3. Insist programmers use a new programming language for these chips and OSes. (This was the case prior to C and Unix.)
4. Insist programmers write in assembly and/or machine code. Perhaps a macro-assembler is acceptable here, but this is inching toward C.
The cost of this flexibility is gross tooling to make it manageable. Can it be done without years and years of accrued M4 and sh? Perhaps, but that's just CMake and CMake is nowhere near as capable as Autotools & friends are when working with legacy platforms.
The justification emerges when you start thinking about how to build a true "package management" system for C libraries which embraces the fact that a world exists beyond GNU/Linux on x86_64.
At some point you realize that not even the (in)famous "host triple" has sufficient coverage for the fractal complexity of the real world.
C and its toolchain are where so many of these tedious little differences are managed. Making tools which account only for the 80% most common use-case defeats the purpose.
How would one encode e.g. "this library targets only the XYZ CPU in Q mode for firmware v5-v9 and it compiles only with GCC v2.95"?
The tools in use need to be flexible enough to handle cases like this with grace or you just end up reinventing GNU Autotools and/or managing your build with a hairy shell script of your own making.
The GNU Autotools system organically grew from dealing with the rapidly-changing hardware and software landscape of the 1980s-1990s. DIY shell scripts were organized, macros were written, and a system was born.
If embedded folk have to start writing their own scripts to handle the inevitable edge cases which WILL come up, then what is a new build tool really accomplishing-- Autotools, but in Python?
That is not a _justification_ - merely an _explanation_ of how we got here. I fully agree that a historical understanding of how we got to where we are is essential for understanding why we have what we have. However, it's critical to not conflate the history and evolution as a justification for why the current system sucks so much! I agree there is a massive amount of complexity and branching to consider, but autotools and the whole C ecosystem does a terrible job of tackling that complexity, and introduces a huge amount of accidental complexity into the solution space.
The co-evolution of hardware, software, and all other moving targets has landed us in a fairly abysmal local maxima. More recently developed toolchains (e.g. zig, rust, etc.) show us that there are much better ways to tackle these problems. Of course they introduce other ones, but we can do so much better.
I like C/C++ a lot, A LOT, and I agree with your comment.
Man, if this got fixed it would be one of the best languages to develop for.
My wishlist:
* Quick compilation times (obv.) or some sort of tool that makes it feel like an interpreted language, at least when you're developing, then do the usual compile step to get an optimized binary.
* A F...... CLEAR AND CONSISTENT WAY TO TELL THE TOOLCHAIN THIS LIBRARY IS HERE AND THIS ONE IS OVER THERE (sorry but, come on ...).
* A single command line argument to output a static binary.
* Anything that gets us closer to the "build-once run-anywhere" philosophy of "Cosmopolitan Libc". Even if an intermediate execution layer is needed. One could say, "oh, but this is C, not Java", but it is already de facto a broken Java, because you still need an execution layer, call it stdlib, GLIB, whatever, if those shared libraries are not on your system with their exact version matching, your program breaks ... Just stop pretending and ship the "C virtual machine", lmao.
The choices are:
1. Restrict the freedom of CPU designers to some approximation of the PDP11. No funky DSP chips. No crazy vector processors.
2. Restrict the freedom of OS designers to some approximation of Unix. No bespoke realtime OSes. No research OSes.
3. Insist programmers use a new programming language for these chips and OSes. (This was the case prior to C and Unix.)
4. Insist programmers write in assembly and/or machine code. Perhaps a macro-assembler is acceptable here, but this is inching toward C.
The cost of this flexibility is gross tooling to make it manageable. Can it be done without years and years of accrued M4 and sh? Perhaps, but that's just CMake and CMake is nowhere near as capable as Autotools & friends are when working with legacy platforms.