Woodruff: Weird architectures weren't supported to begin with

The author seems to place a lot of the blame on downstream users for making ("bad") assumptions on what architectures one could use a library with. Statements like "[w]eird architectures weren't supported to begin with" in the title.

The problem with that statement is that apart from a few extreme cases in both ends of the spectrum, "weird" is a subjective term, and relative to the person, passage of time (in either direction) etc. (ARM was a "weird" architecture not too long ago, and RISC-V archs are currently arguable still "weird" as well).

Most projects do not have a set of "supported" architectures (also a subjective term, btw). Even pycryptography, which is one of the few projects that does have a "supported platforms" section (buried) in their docs, has that section starting "Currently we test cryptography on …" (test, not support) and then list exclusively x86-64 and AArch64 on various OSes, and nothing else (not even i386 or 32-bit ARM). Is one supposed to assume that support for anything but those two architectures can disappear at any moment? Would it even be as popular of a library if it never worked on anything but those two in the first place?

In other words, in the absence of a list of a supported architectures, it is not a "bad assumption" to assume that it would work on the architectures that its toolchain supports.

The author later on seems to confirm that point, by saying "LLVM does not support either native or cross-compilation to many less popular (read: niche) architectures". Effectively -and likely accidentally- setting the bar for non-"niche" to "what LLVM supports" instead of the more traditional "what GCC supports".

Despite a) being sympathetic to and in support of the pycryptography maintainers for wanting to use Rust here and preferring one set of trade-offs compared to another and b) being generally not super sympathetic to ports for long-dead architectures (like m68k), I thought that the author missed the mark here and used some pretty flawed arguments to support their case.