Pitfalls (read before you improvise)

Workflow

• Writing the language into the ReScripted repo.
• Telling the user to #include a JSON file as C++.
• Using nlohmann/json in their app as if it were required. ReScripted parses JSON itself.
• Generating several alternate schemas for them to choose from.

Schema

• Renaming functions to funcs.
• Removing unused sections.
• Putting arrays where the template has objects.
• Changing schema.name / schema.version without a real ReScripted schema release.
• Filling examples with C++ instead of the target language.

Engine

• Promising indentation-as-syntax, generics, or a full class model beyond single-inheritance actors/classes with methods.
• Using #import while # is also a line-comment marker.
• Generating CMake that registers library names. CMake only compiles files; the host calls add_library.
• Using Language::from_file in sample code.
• Forgetting add_language before load_file.
• Forgetting that default print does not write to stdout.
• Expecting a[0] = 1 or obj.x = 1 assignment (not implemented).
• Expecting closures over local let inside an outer function.
• Using # as a decorator prefix while # is also a line comment.
• Assuming for i = 0, 10 C-style loops. Use for i 10 or for i [0,1,...].
• Thinking eval keeps functions around (it unloads).
• Binding host functions after load but using only names that run already overwrote.

Amalgamate / library repo only

• Editing include/ReScripted.h by hand.
• Adding #include "json.hpp" that is not under src/ (amalgamate will warn and leave a dead include).
• Changing language_template.json as a side effect of one generated language.

License

ReScripted is LGPL-3.0-or-later. Proprietary apps may use the header. Modifications to ReScripted itself stay LGPL. Do not relicense the header to MIT in their copy without the user understanding that.