[libc-commits] [libc] [llvm] [libc] Make totalorder functions use the emulated float128 type (PR #221085)

[via libc-commits]

Sukumarsawant wrote:

> @Sukumarsawant
> 
> 1. **Description & AI inquiry**: No AI was used for this PR. The description had broken escaping where backslashes were accidentally used instead of markdown backticks, which introduced literal tab and form-feed characters in the text. I have cleaned up the PR description formatting.
> 2. **File headers**: Added standard Doxygen file headers with `/// \file` per [LLVM Coding Standards](https://llvm.org/docs/CodingStandards.html#file-headers) across all touched implementation and test files.
> 3. **Ordering & Entrypoints**:
>    
>    * Fixed alphabetical ordering of `totalorderf128` and `totalordermagf128` in both `shared_math_test.cpp` and `shared_math_constexpr_test.cpp`.
>    * Added missing `totalorderf128` and `totalordermagf128` entrypoints to `libc/config/darwin/x86_64/entrypoints.txt` and `libc/config/windows/entrypoints.txt`.
> 4. **src/math tests (MPFR)**: `totalorder` and `totalordermag` are IEEE 754 bitwise ordering predicates (checking sign bit, quiet vs signaling, and NaN payloads) rather than numerical/transcendental approximations evaluated against GNU MPFR, so MPFR does not support `totalorder`. Consistent with the other `totalorder` variants in LLVM libc (`totalorder`, `totalorderf`, `totalorderl`, `totalorderf16`, `totalorderbf16`), tests are provided in `libc/test/src/math/smoke/` via `TotalOrderTest.h` and `TotalOrderMagTest.h`.
> 
> All updates have been pushed in the latest commit. Could you please take another look? Thank you!
I won't push on this, but do write comments yourself ( No need to format,etc) . These issues are for exploring the codebase while getting the work done.

Also, the previous comments are not addressed

https://github.com/llvm/llvm-project/pull/221085