Bond is a lightweight bytecode interpreted embeddable scripting language implemented in C++. This is the API of its compiler, virtual machine and runtime library.

Text encoding

All text in Bond is encoded in UTF-8. Every const char * that crosses the C++ API is UTF-8, including:

Bond source code passed to the Lexer, and strings in Bond scripts and CBO files.
File names passed to stream factories, StdioFileHandle and the other I/O classes. On Windows, they are converted to UTF-16 to open the file, so a name that is not valid UTF-8 fails to open. An embedder must not pass names in the ANSI code page.
Names of functions, structs and other symbols, such as those passed to VM::CallFunction.
Command line arguments, which the Bond tools obtain with CommandLineArgs.

A Bond char holds one byte, which is one UTF-8 code unit rather than one character. Compiler error columns count code points.