[armedbear-devel] lambda-list error at compilation
The following works as expected when evaluated, but throws an error when compiling. (defun hashmap-to-hashtable (hashmap &key (keyfun #'identity) (valfun #'identity) (invert? nil) &allow-other-keys &rest rest) (let ((keyset (#"keySet" hashmap)) (table (apply 'make-hash-table (remf :invert? (remf :valfun (remf :keyfun rest)))))) (with-constant-signature ((iterator "iterator" t) (hasnext "hasNext") (next "next")) (loop with iterator = (iterator keyset) while (hasNext iterator) for item = (next iterator) do (if invert? (setf (gethash (funcall valfun (#"get" hashmap item)) table) (funcall keyfun item)) (setf (gethash (funcall keyfun item) table) (funcall valfun (#"get" hashmap item))))) table))) -> Error &REST/&BODY must be followed by a variable. [Condition of type PROGRAM-ERROR] Restarts: 0: [ABORT] Abort compilation. 1: [ABORT] Return to SLIME's top level. 2: [ABORT] Abort thread. Backtrace: 0: (SYSTEM:MAKE-CLOSURE (LAMBDA (#:G29656 &KEY (# NIL #:G29659) (# NIL #:G29661) (# NIL) &ALLOW-OTHER-KEYS ...) NIL) NIL) 1: (JVM::P1-COMPILAND #S(JVM::COMPILAND :NAME HASHMAP-TO-HASHTABLE :LAMBDA-EXPRESSION (LAMBDA (#1=#:G29656 &KEY ((:KEYFUN #:G29658) NIL #:G29659) ((:VALFUN #:G29660) NIL #:G29661) ((:INVERT? #2=#:G29662).. 2: (JVM::P1-FUNCTION #'(SYSTEM:NAMED-LAMBDA HASHMAP-TO-HASHTABLE (HASHMAP &KEY (KEYFUN #'IDENTITY) (VALFUN #'IDENTITY) (INVERT? NIL) &ALLOW-OTHER-KEYS ...) (BLOCK HASHMAP-TO-HASHTABLE (LET # #)))) 3: (JVM::P1 #'(SYSTEM:NAMED-LAMBDA HASHMAP-TO-HASHTABLE (HASHMAP &KEY (KEYFUN #'IDENTITY) (VALFUN #'IDENTITY) (INVERT? NIL) &ALLOW-OTHER-KEYS ...) (BLOCK HASHMAP-TO-HASHTABLE (LET # #)))) Locals: "??" = "??" 4: (JVM::P1 (SYSTEM:NAMED-LAMBDA HASHMAP-TO-HASHTABLE (HASHMAP &KEY (KEYFUN #'IDENTITY) (VALFUN #'IDENTITY) (INVERT? NIL) &ALLOW-OTHER-KEYS ...) (BLOCK HASHMAP-TO-HASHTABLE (LET # #)))) 5: (JVM::P1-LET*-VARS #S(JVM::LET/LET*-NODE :FORM NIL :COMPILAND #S(JVM::COMPILAND :NAME NIL :LAMBDA-EXPRESSION (LAMBDA NIL (SYSTEM:%DEFUN 'HASHMAP-TO-HASHTABLE (SYSTEM:NAMED-LAMBDA HASHMAP-TO-HASHTABLE .. 6: (JVM::P1-LET/LET* (LET* ((#:G29655 #)) (SYSTEM:%DEFUN 'HASHMAP-TO-HASHTABLE #:G29655))) 7: (JVM::P1 (LET* ((#:G29655 #)) (SYSTEM:%DEFUN 'HASHMAP-TO-HASHTABLE #:G29655))) 8: (JVM::P1-FUNCTION-CALL (SYSTEM:%DEFUN 'HASHMAP-TO-HASHTABLE (SYSTEM:NAMED-LAMBDA HASHMAP-TO-HASHTABLE (HASHMAP &KEY # # # &ALLOW-OTHER-KEYS ...) (BLOCK HASHMAP-TO-HASHTABLE #)))) 9: (JVM::P1-BODY ((SYSTEM:%DEFUN 'HASHMAP-TO-HASHTABLE (SYSTEM:NAMED-LAMBDA HASHMAP-TO-HASHTABLE # #)))) 10: (#<FUNCTION {79F502}>) 11: (#<FUNCTION {C49395}>) 12: (#<FUNCTION {E4ED69}> NIL (LAMBDA NIL (DEFUN HASHMAP-TO-HASHTABLE (HASHMAP &KEY # # # &ALLOW-OTHER-KEYS ...) (LET # #)))) 13: (#<FUNCTION (LAMBDA ()) {B5506C}>) 14: (JRUN-EXCEPTION-PROTECTED #<FUNCTION (LAMBDA ()) {B5506C}>) 15: ((FLET SWANK-BACKEND:SWANK-COMPILE-STRING) "(defun hashmap-to-hashtable (hashmap &key (keyfun #'identity) (valfun #'identity) (invert? nil) &allow-other-keys &rest rest) (let ((keyset (#\"..
On May 7, 2010, at 6:53 PM, Alan Ruttenberg wrote:
The following works as expected when evaluated, but throws an error when compiling.
(defun hashmap-to-hashtable (hashmap &key (keyfun #'identity) (valfun #'identity) (invert? nil) &allow-other-keys &rest rest) (let ((keyset (#"keySet" hashmap)) (table (apply 'make-hash-table (remf :invert? (remf :valfun (remf :keyfun rest)))))) (with-constant-signature ((iterator "iterator" t) (hasnext "hasNext") (next "next")) (loop with iterator = (iterator keyset) while (hasNext iterator) for item = (next iterator) do (if invert? (setf (gethash (funcall valfun (#"get" hashmap item)) table) (funcall keyfun item)) (setf (gethash (funcall keyfun item) table) (funcall valfun (#"get" hashmap item))))) table)))
-> Error
&REST/&BODY must be followed by a variable. [Condition of type PROGRAM-ERROR]
Confirmed with a simplified error form not using JSS filed as [ticket #96][1]. [1]: http://trac.common-lisp.net/armedbear/ticket/96 -- "A screaming comes across the sky. It has happened before, but there is nothing to compare to it now."
On Tue, May 11, 2010 at 3:34 PM, Mark Evenson <evenson@panix.com> wrote:
On May 7, 2010, at 6:53 PM, Alan Ruttenberg wrote:
The following works as expected when evaluated, but throws an error when compiling.
(defun hashmap-to-hashtable (hashmap &key (keyfun #'identity) (valfun #'identity) (invert? nil) &allow-other-keys &rest rest) (let ((keyset (#"keySet" hashmap)) (table (apply 'make-hash-table (remf :invert? (remf :valfun (remf :keyfun rest)))))) (with-constant-signature ((iterator "iterator" t) (hasnext "hasNext") (next "next")) (loop with iterator = (iterator keyset) while (hasNext iterator) for item = (next iterator) do (if invert? (setf (gethash (funcall valfun (#"get" hashmap item)) table) (funcall keyfun item)) (setf (gethash (funcall keyfun item) table) (funcall valfun (#"get" hashmap item))))) table)))
-> Error
&REST/&BODY must be followed by a variable. [Condition of type PROGRAM-ERROR]
Confirmed with a simplified error form not using JSS filed as [ticket #96][1].
The error message is misleading, but that lambda list is wrong - &rest must come before &key when both are present.
On May 11, 2010, at 3:50 PM, Alessio Stalla wrote:
Confirmed with a simplified error form not using JSS filed as [ticket #96][1].
The error message is misleading, but that lambda list is wrong - &rest must come before &key when both are present.
Changed the ticket component from "compiler" to "interpreter", which should reject this code. -- "A screaming comes across the sky. It has happened before, but there is nothing to compare to it now."
How about this patch? I'll run the interpreted tests for it shortly, but I guess it makes sense?
On 11 May 2010 19:34, Ville Voutilainen <ville.voutilainen@gmail.com> wrote:
How about this patch? I'll run the interpreted tests for it shortly, but I guess it makes sense?
The interpreted ansi-tests pass without regressions, and the example snippet now errs out with "&REST/&BODY must precede &KEY." I'm not exactly thrilled about the argument processing code, and I bet many people would agree. Rewriting that stuff is another matter and I'm not going to delve into that right now. If you have huge objections to the patch, holler.
participants (4)
-
Alan Ruttenberg -
Alessio Stalla -
Mark Evenson -
Ville Voutilainen