Route hand-rolled diagnostic pragmas through Hedley

[nlohmann]

Fixes #5409.

Stacked on top of #5475 (issue-5408-hedley-undef-leak) since both PRs touch some of the same files. Only the last commit on this branch (Route hand-rolled diagnostic pragmas through Hedley) is new; the base commit is #5475's fix. This PR should be reviewed/merged after #5475 (or rebased onto develop once #5475 lands).

Problem

Several places in the library hand-roll compiler diagnostic suppression with raw #pragma/#ifdef __GNUC__/#ifdef __clang__ guards, instead of using the Hedley primitives already bundled and used elsewhere (JSON_HEDLEY_LIKELY, WARN_UNUSED_RESULT, DEPRECATED_FOR, etc.).

Changes

Converted 6 of the 7 listed push/pop pairs to use JSON_HEDLEY_DIAGNOSTIC_PUSH/JSON_HEDLEY_DIAGNOSTIC_POP and JSON_HEDLEY_PRAGMA(...) instead of raw pragma text:

• include/nlohmann/json.hpp (~3770, ~3863): -Wfloat-equal
• include/nlohmann/detail/conversions/to_chars.hpp (~1078): -Wfloat-equal
• include/nlohmann/detail/output/binary_writer.hpp (~1844): -Wfloat-equal
• include/nlohmann/detail/iterators/iteration_proxy.hpp (~211): -Wmismatched-tags
• include/nlohmann/detail/exceptions.hpp (~36): -Wweak-vtables

iteration_proxy.hpp didn't previously #include <nlohmann/detail/macro_scope.hpp> itself — it only compiled because some earlier header in json.hpp's include list happened to pull it in first. It now includes it directly, like the other detail headers that use Hedley macros, so it's self-contained.

Each pair now calls JSON_HEDLEY_DIAGNOSTIC_PUSH/POP unconditionally (a no-op on compilers that don't need it — that's exactly what the macro is for) and wraps the actual ... diagnostic ignored "..." text in JSON_HEDLEY_PRAGMA(...), going through Hedley's _Pragma()-based emission instead of a raw #pragma line — while keeping the original #ifdef __GNUC__ / #if defined(__clang__) guard around the ignored-pragma itself.

Deviation from the issue's suggested transformation

The issue's example replaces the #ifdef __GNUC__ guard with #if JSON_HEDLEY_HAS_WARNING("-Wfloat-equal"). I did not adopt that part: JSON_HEDLEY_HAS_WARNING is implemented purely via Clang's __has_warning builtin —

#if defined(__has_warning)
    #define JSON_HEDLEY_HAS_WARNING(warning) __has_warning(warning)
#else
    #define JSON_HEDLEY_HAS_WARNING(warning) (0)
#endif

— so it evaluates to 0 on real GCC (which doesn't define __has_warning). Adopting it verbatim would silently stop suppressing -Wfloat-equal on GCC, which is a real regression, not just a style change (the comment even says the guard exists specifically for GCC's behavior around -Wfloat-equal on NaN/0.0 comparisons in constexpr-friendly code). I kept the existing #ifdef __GNUC__ / #if defined(__clang__) guards around the ignored-pragma so the change stays strictly behavior-preserving, and only routed the push/pop/pragma-emission mechanism itself through Hedley, per the issue's actual stated goal ("This must be behavior-preserving").

Two locations intentionally left unconverted

The -Wignored-attributes push at the very top of json.hpp and its matching pop after #include <nlohmann/detail/macro_unscope.hpp> (the workaround for #5103) were not converted:

• The push runs before detail/macro_scope.hpp (and therefore hedley.hpp) has been included anywhere in the translation unit — it's literally the first thing in the file. JSON_HEDLEY_DIAGNOSTIC_PUSH is not yet defined at that point.
• The pop runs after macro_unscope.hpp, which — via hedley_undef.hpp — has already #undef'd every JSON_HEDLEY_* macro (by design, so they don't leak to users, see Four JSON_HEDLEY_* macros leak into user code (PRAGMA, PREDICT_TRUE, PREDICT_FALSE, CLANG_HAS_DECLSPEC_ATTRIBUTE) — not undefined by hedley_undef.hpp #5408/Undefine the four JSON_HEDLEY_* macros that leak after including json.hpp #5475). JSON_HEDLEY_DIAGNOSTIC_POP is no longer defined by the time the pop is reached either.

Making this one pair route through Hedley would require either hoisting the ~2000-line vendored hedley.hpp to the very top of the amalgamated single header (a much bigger structural change to single_include than a pure mechanism swap) or special-casing this one pop ahead of the general macro cleanup at the end of the file. Both are riskier than the mechanical, behavior-preserving change the issue asks for, so I left this one pair as raw pragmas.

Validation

• Compiled include/nlohmann/json.hpp and single_include/nlohmann/json.hpp with -Wall -Wextra -Wfloat-equal -Wmismatched-tags -Wweak-vtables (Clang, which also self-identifies as __GNUC__): no warnings, same as before the change.
• Compiled and ran unit-to_chars.cpp, unit-conversions.cpp, unit-iterators1.cpp, unit-iterators2.cpp, and unit-class_parser.cpp: all pass.
• Compiled unit-msgpack.cpp, unit-bjdata.cpp, and unit-ubjson.cpp (which exercise binary_writer.hpp's write_compact_float extensively): all compile cleanly, and the overwhelming majority of assertions pass (the only failures are pre-existing, unrelated to this change — missing generated test-data files in this offline environment).
• make amalgamate was run; the single_include diff is limited to exactly the lines touched in include/, with no unrelated reordering.
• No real (non-Apple) GCC was available in this sandbox to test directly. The _Pragma("GCC diagnostic ...") text JSON_HEDLEY_PRAGMA emits is byte-identical to the previous raw #pragma GCC diagnostic ... text, and the surrounding #ifdef __GNUC__ guard is unchanged, so GCC's behavior should be identical; CI covers the full GCC matrix.

Breaking change?

No breaking changes. This is purely an internal mechanism change (compiler-diagnostic suppression plumbing); no public API, behavior, or emitted warnings change for any user of the library.

— opened by Claude Code on behalf of @nlohmann