#(foo %)
is just reader shorthand for (fn [arg] (foo arg))
. There's no reason to write a macro that expands into #(...)
. All the %
's in a #(...)
construct are expanded into gensyms right away anyways.
user> `#(foo % %1 %2)
(fn* [user/p1__1877 user/p2__1878]
(user/foo user/p1__1877 user/p1__1877 user/p2__1878))
If you're ever writing a macro that expands to build anonymous functions, you might as well just expand them to fn
forms yourself. In your case, you should probably just use fn
directly and skip the macros. fn
is Clojure's lambda
.
The difference between (fn [] ...)
and (lambda () ...)
in this case is that "fn" is shorter to type than "lambda", and fn
takes a vector for its bindings whereas lambda
takes a list. If you're using Clojure, you will have to get used to this eventually, because vectors are always used for collections of bindings, in all the do
forms and in for
and binding
etc. The rationale behind this as I understand it is that lists are used for function calls or macro calls, and vectors are used for things that aren't calls (lists of symbols to bind, for example). Arguably, it makes it easier to scan the code visually than lists-all-the-way-down. Clojure is not Common Lisp, and you will experience pain if you try to force it to be.
If you really, really wanted to do this, just to say you did:
user> (defmacro lambda [args & body]
`(fn ~(vec args) ~@body))
user> ((lambda (x) (println x)) "foo")
foo
nil
This doesn't let you put a docstring or metadata on your function, among other things. I don't think you would want to use this in a real Clojure program.