LANGUAGE SPECIFICATION

LEXICAL STRUCTURE AND SYNTAX

SPECIAL CHARACTERS

Any UTF-8 byte in the following, not including the period: @$\[]{}";#. They partition naturally into five classes.

• One is sigils: @$.
• The other is brackets: []{}".
• The ; is a singleton partition that is an alias for line termination.
• The next is the singleton #, used for comments.
• The last is the singleton escape: \.

There are situations in which some of the special characters are not special and can be used like a non-special character. They are detailed below, where appropriate.

ESCAPE

The backslash character, \, followed by any character. The character following \ is to be known as the escaped character. A special character is said to be taken literally if it is escaped.

An escape is a static substitution performed at parse time. * If the escaped character is one of the following letters, abfnrtv, it has the standard meaning. * If the escaped character is the asterisk, *, the parser erases all whitespace until the next nonwhitespace character. * Otherwise, the parser ignores any meaning that it would otherwise attribute to the escaped character.

It is not an error to escape a character without meaning. In the string bracket, "", the only escapes that are processed are \* and \" and any other backslash is taken literally. If a quote bracket {} is taken literally, no escapes are processed but escaped {} do not count toward the requirement that all { must be matched with }.

LINE

A sequence of words terminated by newline, \n, or the semicolon, ;. The length of a line is the number of words it contains.

COMMENT

A comment is a special line whose first nonwhitespace character is the pound symbol #. To use # as the first character of a line literally it must be escaped. If it is not the first character of a line, there is no need to escape it. There must be balanced {} in comments. This allows for multiline comments.

WORD

A bracket or UTF-8 character sequence delimited by whitespace or juxtapostion with a bracket. Any word may be prefixed by a sigil. Any special character to be literally included in a word must be escaped. Any UTF-8 character that is not a special character or whitespace may be in the character sequence comprising the word without escaping. A literal word is a word that does not have a sigil and is not a bracket. If the word is not a bracket or sigil, the length of a word is defined as the number of characters the word comprises, after escaping is performed.

SIGIL

The splice, @, or the substitution, $, prefixing a word. The word prefixed by a sigil is said to be the word of that sigil. In a nonprefix position the sigil characters do not need to be escaped to be used as literals. The length of a sigil is defined as the length of the word it prefixes. A sigil must prefix a word.

BRACKET

A sequence of characters between an opening bracket, one of [{" and a closing bracket, one of "}]. A sequence between the [] brackets is a clause. A sequence between the {} brackets is a quote. A sequence between the "" brackets is a word that is delimited by " instead of whitespace, including newlines; no escapes are processed unless the escaped character is " itself or *. The length of a " bracketed word is the length of the characters between "" afer escaping. The only word of length 0 is represented by the literal "". Unbalanced, unescaped brackets are a syntax error.

CLAUSE

A clause is a word that is a line bracketed by []. The length of a clause is the length of the enclosed line.

QUOTE

A word that is a sequence of zero or more lines bracketed by {}. If a quote is used as a command, the usual escaping rules apply. If it is used as a literal word, no escaping is performed. Even if a quote is taken literally, all { must be matched by a properly nested }. Though escapes are not processed, escaped { or } do not count toward nesting. Though it need not be bracketed, the entirety of the input program is a quote. The length of a quote used literally is the length of the characters it contains, not including the outermost {}. A quote is a literal until it is used as code.

EVALUATION

The words in a line are evaluated from left to right, and the line is rewritten to the result of its evaluation.

The lines of a quote are evaluated from top to bottom, and the quote is rewritten to the result of evaluating its last line.

REWRITING

Before a line is evaluated, it's words are rewritten: * A literal word or any quote is written as-is. * A substitution is rewritten as the result of dereferencing its word. * A splice dereferences its word. The result of dereferencing must be a list. Then, the list is expanded in place, increasing the number of arguments by one less the length of the list. If the list is empty, nothing is written, decreasing the number of arguments by one. * A clause is written as the result of executing its enclosed line. If either sigil prefixes the clause, it takes as input the result of rewriting the clause. If a clause is prefixed by a splice, the result of the clause must be a list.

INVOCATION

After rewriting a line is said to be in normal form and the first word in a line is the command and the rest are the arguments of the command. * If the command is a quote or an alien--a command written in Go code--it is invoked with those arguments. * Otherwise, the command is a literal word that is dereferenced. The result of dereferencing must be either a quote or alien. If it is not a quote or alien, evaluation ends in an error.

If the command is a quote, dereferencing the word arguments in its body returns the arguments the callee invoked the quote with.