Comparing Common LISP and Clojure: Cons Cells

Comparing Common LISP and Clojure: Cons Cells

Sarah Sunday

January 24, 2017

The existence (or lack thereof) of cons cells in Clojure might be one of the first things LISPers notice when switching to Clojure. It was for me, anyway. In Common LISP and other LISP languages, cons cells are two element structures that are the basis of all lists and essentially the programming language itself. In LISP, to create a cons cell, simply use cons with the two elements:

? (cons 1 2)
(1 . 2)
? (cons 1 '(2 3 4))
(1 2 3 4)
? (cons '(1 2) '(3 4))
((1 2) 3 4)
? (cons 1 nil)
(1)
? (cons nil 1)
(NIL . 1)
? (cons nil nil)
(NIL)
? (cons nil '(2 3 4))
(NIL 2 3 4)

The two elements can be anything. Two things, two lists, a list and nil, nil and something, and so on. To access the elements of a cons cell, you can do first and rest, if you want to make your code easy to understand for newcomers, but the typical way of doing it is car and cdr.

Which don’t sound like much of anything. CAR? CDR? What do those even mean?

Here’s a quick trip back in time: The early LISP languages were built on old IBM computers that had particular word instructions to access values of a 'register' or machine location. The left half of the register was the address and the right half the decrement. So the LISP way of quickly accessing these values became car and cdr. Respectively, they meant: "Contents of the address part of register" and "Contents of the decrement part of register." And so car and cdr were created to get the first and rest part of a structure. first and rest came into being eventually, but car and cdr stuck around because they were fixtures in the language. Everybody used them. You can't take them away.

But Clojure did. Clojure only has the less-obscurely named and rather dull first and rest to achieve those essential accessing functions.

Yet Clojure did not purge all the oddly named LISP functions. There is a method named cons.

It doesn't make a cons cell, though, not really. It merely takes something (anything) and prepends it to the given sequence, returning it as a list.

user=> (cons 1 2)
IllegalArgumentException Don't know how to create ISeq from: java.lang.Long clojure.lang.RT.seqFrom (RT.java:542)
user=> (cons 1 '(2 3 4))
(1 2 3 4)
user=> (cons '(1 2) '(3 4))
((1 2) 3 4)
user=> (cons 1 nil)
(1)
user=> (cons nil 1)
IllegalArgumentException Don't know how to create ISeq from: java.lang.Long clojure.lang.RT.seqFrom (RT.java:542)
user=> (cons nil nil)
(nil)
user=> (cons nil '(2 3 4))
(nil 2 3 4)

So the one thing you really cannot do with it is use cons with a single element as the second argument. In LISP you can, as you saw in the base example up above that created a dotted list. That doesn’t seem like that jarring of a shift. Just a few use cases! But the entire name of the method is a lie. Clojure isn’t making a true cons cell. The difference in the meaning of the function cons is vastly different. Cons cells aren't the underlying implementation of lists in Clojure. Cons cells have nothing to do with cons.

Clojure—by keeping the name and not the substance—is programmatically appropriating from Common LISP for reasons I don't really understand.

It really isn't the end of the world, though, as much as it doesn't make sense. There are far worse things to do in a programming language than keep a bizarre name for something because, well, that's how it's done.

...

I suppose Common LISP and Clojure aren't all that different.