Re: Functions named by keywords

Re: Functions named by keywords

Keywords are constants in the variable namespace; binding them as functions binds them in the function namespace. You can try it in any other case, e.g. (DEFCONSTANT FOO 42) and then (DEFUN FOO () 24) and then (LIST FOO (FOO)).

I’m maybe remembering it wrong, but keyword should resolve to itself, defining it as a function will break this rule. It might break things i do not know about so i would avoid it and follow the rules. 

I’m just light user anyway so my response might be completely wrong. If this is the case, then please correct me.

Ladislav Koščo 

+421-949-49-36-36

On 25 Oct 2020, at 09:50, Rainer Joswig <joswig@lisp.de> wrote:

 for LispWorks see the documentation for system:define-top-loop-command

CL-USER 4 > (system:define-top-loop-command (:reverse-it (:result-type values))

                (&rest args)

              (reverse args))

(:REVERSE-IT)

CL-USER 5 > :reverse-it 1 2 3 4

(4 3 2 1)

Am 25.10.2020 um 09:19 schrieb Pascal Bourguignon <pjb@informatimago.com>:

On 24 Oct 2020, at 19:12, Marco Antoniotti <marco.antoniotti@unimib.it> wrote:

I agree with Pascal.  It is one of the consequences of CL being a 2-Lisp.

Having said that, please DO NOT DO IT!  IMHO, there are already way too many abuses of keywords when a proper symbol would be much more appropriate.

I think we may allow the end-user to fbind keywords for interactive commands.  

This let you implement REPL commands that are easily accessible from any package.

But of course, this is not something that you would do in library code, eg. code published on quicklisp.

Note that some CL (ccl, ecl) debuggers use keywords as commands (including function forms), because this allows them to read them using CL:READ whatever the current *PACKAGE*.

-- 
__Pascal J. Bourguignon__

Re: Functions named by keywords

[glossing over details ... ]

In CL, symbols are 2-tuples: {package, name}.

CL is a lisp2.  Symbols have a value and a function and ...

The Reader understands 2-tuple symbols {abc, xyz} when written with a double colon, e.g. abc::xyz.

When the compiler/interpreter sees a list, it interprets it as (<function> <val> <val> ...).  This is still just a list, it is the *compiler* which interprets the *meaning* of the list as "something more than a list".

When a symbol is in the <function> position, the compiler looks up the function value of the symbol.

When a symbol is in the <val> position, the compiler looks up the value of the symbol (which can be different from the function value of the symbol).

CL has a built-in shorthand for KEYWORD::xxx symbols.  The shorthand uses a single colon (not a double colon).  E.G. the reader expands :xxx to be KEYWORD::xxx (the 2-tuple {“KEYWORD", “xxx"}).

CL has another built-in shorthand for xxx symbols (i.e. no package name given, no colons at all).  The reader expands xxx to be <package>::xxx [where <package> is the current package, defined as the value of *package*].

DEFUN and DEFMETHOD take a bunch of args, the first one being a symbol (i.e. a 2-tuple symbol).

Usually, one uses the form (DEFUN xxx ...).  The reader expands this to mean (DEFUN <package>::xxx ...).

But, furthermore... E.G. I have added methods for classes by using:

(defmethod abc::xyz (...) ...)  [where abc is the package that the class symbol is in, and xyz is some new function/method that I want to add to the class]

So, it would not surprise me that keywords can be used in DEFUN and DEFMETHOD.  (In fact, DEFUN and DEFMETHOD are just 2-tuple symbols, too, but I don't want to go there :-).

Many non-CL Lisps are lisp1's, and use 1-tuple symbols (name only).  Many other languages don't have the concept of symbols and use various syntactic kludges to attack the packge problem.

pt


_______________________________________________
Lisp Hug - the mailing list for LispWorks users
lisp-hug@lispworks.com
http://www.lispworks.com/support/lisp-hug.html

Re: Functions named by keywords