A set of procedures that implement the basics of functional programming – as in languages like LISP and Haskell – in non-object-oriented REXX.
In particular, it provides the following features:
- Anonymous, first-class functions (called using a special
?procedure)- Basically, procedures can be stored in variables
- This allows for higher-order procedures: procedures whose arguments or return values are themselves procedures
- Simple partial application (currying)
- This makes it easy to make simple procedures from more complex ones
car,cdr, andconsfrom LISP- A pair of values can be stored in a single variable
- This allows for the creation of more complicated data types (especially linked lists)
- Conversion between REXX stem variables and LISP-style linked lists
- This provides somewhat greater interoperability
- Lexical scope (kind of)
However, attempting to do this in REXX causes several issues:
- Performance will likely be an issue, since the library uses
INTERPRETto perform function calls - REXX's relatively low call stack limit (100) prevents the use of recursion for most tasks
- As REXX lacks true compound data types, all data must be stored in strings (ugh)
- Lexical scope is possible, but it is hardly elegant
In general, this is probably more of a hack than something of great practical value. Still, it is interesting to see how functional programming can be implemented in a language like REXX.
In the first place, REXX is frequently used to complete the sorts of data processing tasks for which functional programming can be suitable. Secondly, this provides a way of allowing more powerful REXX programs without needing to implement OOP (which is unavailable on some REXX platforms).
In the second place, why not?
- Ramda, a functional programming library for JavaScript
- Scheme, a LISP dialect
- PHP's closure syntax