This is a sort of formal specification for Jux, to be used to actually create compilers and interpreters for it later.
Jux has a very simple syntax. Tokenizing also is very straightforward:
- EOL comments start with
# - Multiline comments are delimited with
(*and*). Multiline comments cannot be nested. (TODO: Good choice?) - Strings are delimited with
".\"can be used to escape a double quote. - All other kinds of tokens are 'words'.
- Words are separated with one or multiple whitespaces.
[is always a word on its own.]is always a word on its own (regardless of whitespace).- Words are case-sensitive.
- A word that starts with
\, such as\foois called an 'atom'.
A special word, def, reads three arguments from the stack:
- the name of the to-be-defined word, as atom.
- A documentation string.
- A quotation which is the function body/implementation.
The current function dictionary is then taken and updated with the new function on top. The function dictionary field should contain:
- the function name (as key)
- the function's body, where for each of the contained words, their implementation has been looked up in the function dictionary (before this new word was added).
- For quotations, this happens as well for the contained elements; each element of the quotation being a
(word, implementation)tuple.
The Jux program should have a method called main, which is the starting point of the program.
TODO: Modules.