Rapid YAML Files

Breaking changes

• [BREAKING] Fix #480 (PR#489):
• Deserializing an empty quoted string will not cause an error.
• Deserializing an empty string will cause an error: the empty string is read in as an empty scalar.
• Ensure keys are deserialized using all the rules applying to vals.
• Added KEYNIL and VALNIL to NodeType_e, used by the parser to mark the key or val as empty. This changed the values of the NodeType_e enumeration.
• Added NodeType::key_is_null() and NodeType::val_is_null().
• [BREAKING] Fix #477 (PR#479): changed read<std::map>() to overwrite existing entries. The provided implementations had an inconsistency between std::map (which wasn't overwriting) and std::vector (which was overwriting).

Fixes

• PR#488:
• add workarounds for problems with codegen of gcc 11,12,13.
• improve CI coverage of gcc and clang optimization levels.
• PR#496 and c4core PR#148: Add CI-proven support for CPU architectures:
• mips, mipsel, mips64, mips64el
• sparc, sparc64
• riscv64
• loongarch64
• Fix #476 (PR#493): add handling of Byte Order Marks.
• PR#492: fix emit of explicit keys when indented: yaml fixed: ? explicit key : value previously: ? explicit key : value # this was not indented
• PR#492: fix parser reset for full reuse (m_doc_empty was not resetted), which would cause problems under specific scenarios in subsequent reuse.
• PR#485: improve the CI workflows (thanks to @ingydotnet):
• amazing code reuse and organization, thanks to the use of YamlScript to generate the final workflows
• all optimization levels are now covered for gcc, clang and Visual Studio.
• PR#499: fix warnings with -Wundef.

Thanks

• @ingydotnet
• @perlpunk
• @Delian0