On Tue, Dec 7, 2010 at 2:15 AM, javajosh <javaj...@gmail.com> wrote: > Mike and I have had a nice off-line conversation where we enumerated > the possible things that can come after open-parens. I listed 7, he > added 3: > >> 1. A value (if the paren has a tick '( ) >> 2. A function. >> 3. A map - which is a psuedo function that takes a key as an arg. >> 4. A keyword - which is a psuedo function that takes a map as an arg. >> 5. A macro. This is the normal case, I think. Looking through the mailing >> list, it appears that most clojure programming questions revolve around >> which one of the hundreds of macros to invoke and in which order! >> 6. The Java form: (MyClass. ) >> 7. The java method form (.doSomething) > 8. A function returning a function to invoke - ((find-my-function) ) > 9. A loop - (recur ) > 10. The anonymous function macro: #( ) > > So, at least I know why I feel uneasy about open paren! It's super > overloaded.
Not really. (...) is a non-atomic s-expression. If it's evaluated unquoted, the first nested s-expression is evaluated and if it's not callable an exception is thrown. Macros, special forms (which are sort of like system-internal macros and are used to build all the other macros, and functions), Java constructors, Java methods, and functions are callable (and maps and keywords -- also vectors -- act as functions for this purpose). The only real overloading always involves macros: #() evaluates to a function and doesn't run its insides right away. Then again so does (defn foo [x] (inc x)) -- the (inc x) is run when foo is called, calling inc, but not when the (defn ...) is called. Macros can delay evaluation of their contents, and #() is a reader macro. '(x y z) is another reader macro and expands to (quote (x y z)). And lastly some macros use a Common Lisp style of expecting a parenthesized list argument whose first element isn't (at some point) called as an invokable of some sort. The ns macro is particularly guilty of this. I say guilty because I think it's bad design, which I guess may have been grandfathered in before the standard was settled on to use [] around non-executable lists of data such as binding lists. (In some cases the first element, e.g. when it's :use or :require, can be thought of as a pseudo-operator defined solely within the ns form. I don't have an objection in cases where the first element of the list can be thought of as a command or a call of some sort. But many of the nested inputs to e.g. :use have the stain.) But (x y z w) is similar to C's x(y z w); or Java's y.x(z w); in what it generally means. Except when quoted, or in a few macro arguments (particularly, ns arguments). > Mike also points out that things that aren't functions (not used in > that context) can't be aliased with def or use. Really? user=> (def a 3) #'user/a user=> (def b a) #'user/b user=> b 3 -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en
