- Tcl has 1 (one) data type: everything is a string; Pils has 2 (two) data types: the atom and the list.
- The procedure string has the new sub command append. (Btw, by formal reasons, it would be nice to see it also in regular Tcl. I think that it is not really expensive!)
- The list is not cons'd but instead a simple Tcl list. (This idea is stolen from NewLisp [1].)
- No cons, car, cdr, instead lindex etc. No define but instead proc. (*1)
- No symbols as every atom can serve as such by saying (set varName) or using the shortcut $varName. (I told you. As Tcl-ish as possible.)
- for and foreach not yet implemented
- many many ... hrmpf ... bux?
(*1) RS: When you're in the process of making a Lisp-like language, I think the unification in Scheme
(define var 'value) (define (function x y) (+ x y))is a good idea (and can of course easily be had in Tcl, see the Scheme page). The only trouble is that argument-less functions can not be expressed this way...wdb: This leads to the consequence that procs are referenced by vars as in Scheme. My attempt was to make it extremely Tcl-ish. In Tcl, proc names are not variable names but can be thought of as pointers. Of course, you can always say:
(set x list) ;; line 1 ($x a b c) ;; line 2 ;; equivalent to (list a b c), note the shortcut $Here as well as in Tcl, list is used as a pointer to a procedure, x is used as a pointer to a string constant. In line 2, we see a pointer to a pointer to a procedure. That wasn't my intention, but instead to combine the minimalism of Tcl with the ease of Lisp. Take it as base for a discussion with open result.But, argument-less function could be easily made with a macro define as Pils recognises the type of its data. A list containing atom x == {0 {{1 x}}} does not equal the atom x == {1 x} itself. (That's the difference to Tcl.)And, thank you for your comment. This is what I wanted.
