Let me share my sponaneous thoughts.
On Fri, 07 Nov 2025 11:04:50 +0000
whistlez <[email protected]> wrote with subject
"Fil-C - Memory Safe C/C++ Compiler":
> I'd like your opinion on this project and whether it could be useful in
> an OpenBSD environment.
(Part I - not related to OpenBSD:) I have two somewhat unusual C/asm projects,
one using x86 assembler string opcodes and mmaps for a kind of ringbuffer and
the other using mangled pointers (I try to adapt a unique-pointer-like workflow
known from C++ in C, but try to be not so strict and to avoid cumbersome
keywords/constructs). When I pick them up again, which can take quite a while,
maybe then I will check them against this Fil-C compiler.
(Part II - related to OpenBSD:) I bet this compiler will not as important part
of OpenBSD. Reasons:
* Required garbage collector may conflict basic design principles?
* OpenBSD code probably needs to be aligned to this compiler, making it a
de-facto standard (because if you use GCC, you must respect those specific
"capabilities" like InvisiCaps, even if it's just in a fraction of the code)
* There will be performance implications (homepage says up to factor 4 in bad
cases)
> Naturally, this compiler's collator effects degrade executable
> performance, but that doesn't concern us.
<- Performance matters. One would need reliable numbers for the performance
degration, but in former discussions on this list I have seen people arguing
that it is very hard not to forget important use cases when gathering
performance numbers.
