Woodruff: Weird architectures weren't supported to begin with

The Foobar9000 does some things that probably made sense at the time (segmented pointers, sign-magnitude integers, the IA64 NaT bit, putting something interesting at address 0, etc.), but my code is sufficiently low-level (e.g. a GCC or LLVM backend) that dealing with those things is actually painful and I don't feel like bothering.

The Foobar9000 does not have (an MMU/software programmable interrupts/an FPU/etc.) and I'm not willing to (replace all calls to fork() with vfork()/insert gratuitous sched_yield()s everywhere/avoid using floats/etc.).

I've never heard of anyone running any sort of software on a Foobar9000 without using FooCorp's proprietary non-Unix operating system, and I'm not porting my software to that because I would basically have to rewrite the whole thing from scratch.

I've never heard of anyone running any sort of "real" operating system on a Foobar9000 at all (e.g. because it was purpose-built to do exactly one thing, like the Apollo Guidance Computer), and my code is obviously incapable of running on the bare metal.