blog | oilshell.org

Three Compilers to Make a Lexer

Project Background

Maybe put this in the next post. Yeah because that's where I talk about re2c.

Ninja.

C++ project.

• The lexer is what convinced me I could finish Oil.
• The lexer contains a lot of complexity.
  • 14 modes
  • 233 IDs.

Future Work

• Measuring Speed
• Rewriting more of the lexer in C. Not just the pattern matching parts.
• BASH_REGEX_CHARS state
• Oil states -- adding the oil language to OSH.
• Reducing size by subdiving OUTER into its components
  • this will reduce speed, so it will be good to measure it.