# Xah Language Design

programing some quick notes of thoughts.

been studying computer languages for years, especially past 2. In past year, i thought of designing my own lang. Here's some sketch.

the most important element of a computer language is its syntax. (John Backus shares similar view)

• each symbol must mean one thing only. No symbol is used for more than one purpose. e.g. each symbol doesn't depend on adjacent symbols for its meaning.• make use of unicode. It's there.

• Overall, the syntax will be like lisp sexp but with also m-expression. e.g. similar to Wolfram Language with FullForm. (What Mathematica calls FullForm i'll call tentatively Regular Nested Form.)

• the syntax will be completely regular, with a grammar. Mathematica is close to this, no other lang is i know of has regular grammar. Note: lisp syntax has many irregularities

So, to begin with, the Regular Nested Syntax will be like this:

- ① 「」
- ② 「atom」
- ③ 「atom1, atom2, …」
- ④ 「expr」

and the “expr” can be any of ① ② ③.

The corner bracket 「」 is similar to lisp's paren.

For example, lisp's`(* 5 (+ 3 4))`

would be `*, 5, 「+,3,4」`

.
comma is used as separator, not space. Space (U+20) has no meaning whatsoever. Space is used only for visual display or programer's habit of typing it. e.g. 「3 4」 doesn't mean 3*4 as in Mathematica. (there will be painless syntax transformation facilities built into the editor, similar to Mathematica's FrontEnd.)

the m-expression will be spec out later, because there's a lot. (think of Mathematica for now) Each function, such as Map, MapThread, Join, Part, etc will have a operator.

first, few things about matching pair symbols.

- • the paren () is used grouping order of operation only.
- • the square bracket [] is for function argument.
- • the curly bracket {} is syntax sugar for list.
- • i think that each operator will be one single unicode symbol. For example, there wouldn't be ==, !=, <=, :=, ++, :: etc. All operators will be single unicode symbol, except matching pairs etc.

• The asterisk 「*」 will be the multiplication operator. Not ×. Because the traditional diagonal cross is too similar to letter x.

the Editor will have built-in tools (example: a button) that changes syntax on the fly. 2 form at least. ① convert any sourc code to regular nested form. ② another standard form for normal reading (similar to Mathematica StandardForm). Any source code can trivially be formatted in one of these forms (by pressing a button)

…

a little on semantics.

• the lang will be somewhat like Mathematica in that everything should be builtin as much as possible, including all mathematical operations such as integral, derivative, solving equations, like Mathematica.

• possibly, the lang will forbid setting variables. i.e. like Haskell.

• the lang will of course be functional. In fact, consider offering no procedural operation at all.

• the lang will contain no macros (as in lisp). But *possibly* pattern matching (as powerful as Mathematica), but am not sure about pattern matching...

• possibly consider John Backus's Functional-Level paradigm.

• possibly consider a complete point-free syntax.

• there'd be no call-cc or such.

• the lang will probably be dynamically typed, and the types will be mathematical, like Mathematica. e.g. Integers, Real Numbers, Complex Numbers, Roots. No shits such as int, long, float, double.

- Math Notation, Computer Language Syntax, and the “Form” in Formalism
- Programing Language Syntax Soup

If you have a question, put $5 at patreon and message me.