Comment by pezezin
3 days ago
Portable to what? Rust works fine on all of today's popular platforms.
I see people complaining about Rust's lack of portability, and it is always some obsolete platform that has been dead for 20 years. Let's be serious, nobody is gonna run Tor on an old SGI workstation or Itanium server.
>Let's be serious, nobody is gonna run Tor on an old SGI workstation or Itanium server.
dont temp me with a good time and awesome weekend project!
It is still possible to build for targets such as Itanium. There is nothing stopping you from writing your own targets too in LLVM and its not obscenely difficult
https://github.com/llvm/llvm-project/tree/main/llvm/lib/Targ...
Not under OpenBSD i686.
Also, the more heterogeneous it's a CPU architecture support, the least exploitable a service will be.
Please elaborate on your last point.
See? This is exactly my point. The last 32-bit only Intel CPUs were the original Intel Core series. Core 2 was already 64-bit, and it was released in 2006.
i686 should have died a long, long time ago.
No, the last 32-bit-only Intel CPUs were Atom models in ... I think we decided 2011 last time this came up.
That's a good point, really there's no reason to waste time on anything but popular platforms. Obviously, of course, this means dropping support for everything except Windows (x64 and maybe ARM) and macOS (ARM). (\s)
In all seriousness, I guess you can make this argument if you only care about Windows/macOS, but the moment you run anything else I have to ask why, say, Linux deserves support but not other less-common platforms.
Because:
1) Development resources are finite.
2) Linux runs all of the world supercomputers, most of the internet infrastructure (server, routers, etc), most of the cellphones (Android), and lots of other things. Its global marketshare is way bigger than macOS and all the BSD put together.
To remain clear: This is largely devil's advocate. I believe that niche platforms generally should be supported, and that includes GNU/Linux on amd64, and NetBSD on literal VAX (yes, that is an officially supported platform in current NetBSD), and RedoxOS on ARM, and OpenBSD on MIPS, and [...]. I just think it's really weird to claim that GNU/Linux, with 3% of the desktop market, should be supported, but a platform that is similarly a small fraction of Linux is dead weight that should be dropped.
With that said, remainder of this comment continues with the position that GNU/Linux, which I am writing this comment on, is obviously not worth supporting for the same reasons as i.e. MIPS and RISC-V.
> 1) Development resources are finite.
That is an argument in favor of cutting niche platforms like GNU/Linux.
> 2) Linux runs all of the world supercomputers,
You can't defend a niche OS by pointing out that it's used in a tiny niche market. How many supercomputers exist on earth? I'd bet you there are more working MIPS installations than supercomputers.
> most of the internet infrastructure (server, routers, etc),
I'll grant you headless machines used by IT folks, but that's still a specific subset of the market and it has little bearing on whether, say, Tor should support it as a desktop OS.
> most of the cellphones (Android),
Android/Linux is quite different from GNU/Linux; effectively nobody developing for Android is targeting Linux in any meaningful sense.
> and lots of other things. Its global marketshare is way bigger than macOS and all the BSD put together.
Only if we include embedded systems and servers. If you intend to target servers, then yes obviously Linux matters. Otherwise, not so much.
Try living without OpenSSH and the rest of *BSD contributions.
Linux is not an ISA
So? They're both part of the platform software has to target.