@stereo I can't even build the libc crate on 32-bit PowerPC, something's defined two times, obviously the preprocessor condition is broken somewhere, but I'm too lazy to fix.
But 32-bit PowerPC on Linux MUSL is considered a tier 3 arch if not further below so no surprises here. I'm not sure if RISCV support is any good, but it's more "current" and still alive, hopfully there's more interest in fixing issues with that 😅
@m0xee maybe i cant build it either. its still "ongoing" ;)
he other server is using golang which was building fine. i also cross-compiled molly on my x86 desktop
@stereo Try gmid — this one should be pure C and it worked better than Molly for me, with Molly I ran into some issues with IPv4 being a port redirect from other host and having a native IPv6 address (over a tunnel, it's complicated 🤣)
I ended up using an old ARM machine for my Gemini stuff with gmid, and PPC one for the rest, like my Pleroma instance (erlang still works) and few others.
I'm a huge fan of marginal CPU architectures, I wish I could have a RISCV machine to experiment on 😄
@stereo Unfortunately, Go on PPC is a no go — there is no reference toolchain, and although I have managed to build gcc-go, most modules don't support this arch. Rust works, but, of course, with a lot of caveats. I'm in the process of getting libreddit work, it has ring crate among its dependencies — cryptography is of course tricky with lots of low-level stuff, I have managed to make IBM's port to ppc64 build, but even tests don't pass, something's to do with endian-ness I guess 😩
@stereo I have managed to make Nitter work though. Mostly by using native libraries, nim modules had issues with bit shift operations not working the way they are supposed to. That was a cheat, but that's the main difference between legacy architectures and current ones — with former ones you can exploit the past porting efforts, with current ones you can do a lot of work yourself, but can get help from community 😄