Jacob Young

Crashing instead of producing an error is a compiler bug, but note that `@typeInfo(@This())` contains the values of all of the enum tags, which requires resolving the value of `U`...

Just tacked on handling and testing of more cases.

Note that `std.mem.readPackedIntLittle(T, bytes, bit_offset)` is the same as `std.mem.readPackedInt(T, bytes, bit_offset, .Little)` and `std.mem.readPackedIntBig(T, bytes, bit_offset)` is the same as `std.mem.readPackedInt(T, bytes, bit_offset, .Big)`.

This merge has been reverted in 43c98dc11567eeb38be041c7dad179c53156f3df while the large variety of random failures that occur with the build system on my local configuration in response to this change are...

The previous revert has been unreverted in 2282c278850e29af81e463d55c81fa08855823f8 due to finally diagnosing the issue as being related to linux kernel 6.4.0 and not this PR.

I realize that this accepted definition of `@expect` matches an llvm intrinsic, but I do not believe that this is the most helpful tool for the programmer, for self-hosted backends,...

This was changed in 73c39adda099528d5776545a6d3b4daa977de3e8 because all of the variables in the label file were showing up as having a value of 0.

I'm not even sure what `zig test-compare` was referring to. The closest thing I can find is `zig build test-compare-output` which works fine on master.

> But zig.h is still not getting copied to the output dir. Covered by #13528

This backend has always has always supported avoiding undocumented instructions, but there's not any frontend support for it yet. The simplest way to pass it to clang is something like...