Re: possibilities for macros in a typed language

Taking this a step further...

Let's start with Kitten (Cat minus locals). Now let's also remove  
recursive definitions. In such a language, any definition is  
equivalent to its finite expansion. In other words, for every place  
the function 'f' is used, replacing 'f' with its definition will not  
alter the meaning of the program.

If we do this, writing a program would consist of writing a series of  
rewrite rules. For example, if we do 'sqr = dup *', then every time  
'sqr' is encountered, we can replace it with 'dup *'. (If the  
implementation actually does this or decides to call 'sqr' instead  
isn't interesting; the results are equivalent.)

But why stop there? What if we allow rewrite rules along the lines of  
MetaCat as normal definitions? Even though we can't define 'm = dup  
i', we could define '$a m = $a $a i'. Optimizations like 'swap swap =  
' could sit aside normal definitions.

I feel like I'm probably reinventing something here. Are there any  
existing languages that work like this?

- John