;;;; miscellaneous side-effectful tests of CLOS ;;;; This software is part of the SBCL system. See the README file for ;;;; more information. ;;;; ;;;; While most of SBCL is derived from the CMU CL system, the test ;;;; files (like this one) were written from scratch after the fork ;;;; from CMU CL. ;;;; ;;;; This software is in the public domain and is provided with ;;;; absolutely no warranty. See the COPYING and CREDITS files for ;;;; more information. (load "assertoid.lisp") (defpackage "CLOS-IMPURE" (:use "CL" "ASSERTOID")) (in-package "CLOS-IMPURE") ;;; It should be possible to do DEFGENERIC and DEFMETHOD referring to ;;; structure types defined earlier in the file. (defstruct struct-a x y) (defstruct struct-b x y z) (defmethod wiggle ((a struct-a)) (+ (struct-a-x a) (struct-a-y a))) (defgeneric jiggle (arg)) (defmethod jiggle ((a struct-a)) (- (struct-a-x a) (struct-a-y a))) (defmethod jiggle ((b struct-b)) (- (struct-b-x b) (struct-b-y b) (struct-b-z b))) (assert (= (wiggle (make-struct-a :x 6 :y 5)) (jiggle (make-struct-b :x 19 :y 6 :z 2)))) ;;; Compiling DEFGENERIC should prevent "undefined function" style ;;; warnings from code within the same file. (defgeneric gf-defined-in-this-file (x y)) (defun function-using-gf-defined-in-this-file (x y n) (unless (minusp n) (gf-defined-in-this-file x y))) ;;; Until Martin Atzmueller ported Pierre Mai's CMU CL fixes in ;;; sbcl-0.6.12.25, the implementation of NO-APPLICABLE-METHOD was ;;; broken in such a way that the code here would signal an error. (defgeneric zut-n-a-m (a b c)) (defmethod no-applicable-method ((zut-n-a-m (eql #'zut-n-a-m)) &rest args) (format t "~&No applicable method for ZUT-N-A-M ~S, yet.~%" args)) (zut-n-a-m 1 2 3) ;;; bug reported and fixed by Alexey Dejneka sbcl-devel 2001-09-10: ;;; This DEFGENERIC shouldn't cause an error. (defgeneric ad-gf (a) (:method :around (x) x)) ;;; DEFGENERIC and DEFMETHOD shouldn't accept &REST when it's not ;;; followed by a variable: ;;; e.g. (DEFMETHOD FOO ((X T) &REST) NIL) should signal an error. (eval-when (:load-toplevel :compile-toplevel :execute) (defmacro expect-error (&body body) `(multiple-value-bind (res condition) (ignore-errors (progn ,@body)) (declare (ignore res)) (typep condition 'error)))) (assert (expect-error (macroexpand-1 '(defmethod foo0 ((x t) &rest) nil)))) (assert (expect-error (defgeneric foo1 (x &rest)))) (assert (expect-error (defgeneric foo2 (x a &rest)))) (defgeneric foo3 (x &rest y)) (defmethod foo3 ((x t) &rest y) nil) (defmethod foo4 ((x t) &key y &rest z) nil) (defgeneric foo4 (x &rest z &key y)) (assert (expect-error (defgeneric foo5 (x &rest)))) (assert (expect-error (macroexpand-1 '(defmethod foo6 (x &rest))))) ;;; more lambda-list checking ;;; ;;; DEFGENERIC lambda lists are subject to various limitations, as per ;;; section 3.4.2 of the ANSI spec. Since Alexey Dejneka's patch for ;;; bug 191-b ca. sbcl-0.7.22, these limitations should be enforced. (labels ((coerce-to-boolean (x) (if x t nil)) (%like-or-dislike (expr expected-failure-p) (declare (type boolean expected-failure-p)) (format t "~&trying ~S~%" expr) (multiple-value-bind (fun warnings-p failure-p) (compile nil `(lambda () ,expr)) (declare (ignore fun)) ;; In principle the constraint on WARNINGS-P below seems ;; reasonable, but in practice we get warnings about ;; undefined functions from the DEFGENERICs, apparently ;; because the DECLAIMs which ordinarily prevent such ;; warnings don't take effect because EVAL-WHEN ;; (:COMPILE-TOPLEVEL) loses its magic when compiled ;; within a LAMBDA. So maybe we can't test WARNINGS-P ;; after all? ;;(unless expected-failure-p ;; (assert (not warnings-p))) (assert (eq (coerce-to-boolean failure-p) expected-failure-p)))) (like (expr) (%like-or-dislike expr nil)) (dislike (expr) (%like-or-dislike expr t))) ;; basic sanity (dislike '(defgeneric gf-for-ll-test-0 ("a" #p"b"))) (like '(defgeneric gf-for-ll-test-1 ())) (like '(defgeneric gf-for-ll-test-2 (x))) ;; forbidden default or supplied-p for &OPTIONAL or &KEY arguments (dislike '(defgeneric gf-for-ll-test-3 (x &optional (y 0)))) (like '(defgeneric gf-for-ll-test-4 (x &optional y))) (dislike '(defgeneric gf-for-ll-test-5 (x y &key (z :z z-p)))) (like '(defgeneric gf-for-ll-test-6 (x y &key z))) (dislike '(defgeneric gf-for-ll-test-7 (x &optional (y 0) &key z))) (like '(defgeneric gf-for-ll-test-8 (x &optional y &key z))) (dislike '(defgeneric gf-for-ll-test-9 (x &optional y &key (z :z)))) (like '(defgeneric gf-for-ll-test-10 (x &optional y &key z))) (dislike '(defgeneric gf-for-ll-test-11 (&optional &key (k :k k-p)))) (like '(defgeneric gf-for-ll-test-12 (&optional &key k))) ;; forbidden &AUX (dislike '(defgeneric gf-for-ll-test-13 (x y z &optional a &aux g h))) (like '(defgeneric gf-for-ll-test-14 (x y z &optional a))) (dislike '(defgeneric gf-for-ll-test-bare-aux-1 (x &aux))) (like '(defgeneric gf-for-ll-test-bare-aux-2 (x))) ;; also can't use bogoDEFMETHODish type-qualifier-ish decorations ;; on required arguments (dislike '(defgeneric gf-for-11-test-15 ((arg t)))) (like '(defgeneric gf-for-11-test-16 (arg)))) ;;; structure-class tests setup (defclass structure-class-foo1 () () (:metaclass cl:structure-class)) (defclass structure-class-foo2 (structure-class-foo1) () (:metaclass cl:structure-class)) ;;; standard-class tests setup (defclass standard-class-foo1 () () (:metaclass cl:standard-class)) (defclass standard-class-foo2 (standard-class-foo1) () (:metaclass cl:standard-class)) (assert (typep (class-of (make-instance 'structure-class-foo1)) 'structure-class)) (assert (typep (make-instance 'structure-class-foo1) 'structure-class-foo1)) (assert (typep (make-instance 'standard-class-foo1) 'standard-class-foo1)) ;;; DEFGENERIC's blow-away-old-methods behavior is specified to have ;;; special hacks to distinguish between defined-with-DEFGENERIC-:METHOD ;;; methods and defined-with-DEFMETHOD methods, so that reLOADing ;;; DEFGENERIC-containing files does the right thing instead of ;;; randomly slicing your generic functions. (APD made this work ;;; in sbcl-0.7.0.2.) (defgeneric born-to-be-redefined (x) (:method ((x integer)) 'integer)) (defmethod born-to-be-redefined ((x real)) 'real) (assert (eq (born-to-be-redefined 1) 'integer)) (defgeneric born-to-be-redefined (x)) (assert (eq (born-to-be-redefined 1) 'real)) ; failed until sbcl-0.7.0.2 (defgeneric born-to-be-redefined (x) (:method ((x integer)) 'integer)) (defmethod born-to-be-redefined ((x integer)) 'int) (assert (eq (born-to-be-redefined 1) 'int)) (defgeneric born-to-be-redefined (x)) (assert (eq (born-to-be-redefined 1) 'int)) ;;; In the removal of ITERATE from SB-PCL, a bug was introduced ;;; preventing forward-references and also change-class (which ;;; forward-references used interally) from working properly. One ;;; symptom was reported by Brian Spilsbury (sbcl-devel 2002-04-08), ;;; and another on IRC by Dan Barlow simultaneously. Better check ;;; that it doesn't happen again. ;;; ;;; First, the forward references: (defclass a (b) ()) (defclass b () ()) ;;; Then change-class (defclass class-with-slots () ((a-slot :initarg :a-slot :accessor a-slot) (b-slot :initarg :b-slot :accessor b-slot) (c-slot :initarg :c-slot :accessor c-slot))) (let ((foo (make-instance 'class-with-slots :a-slot 1 :b-slot 2 :c-slot 3))) (let ((bar (change-class foo 'class-with-slots))) (assert (= (a-slot bar) 1)) (assert (= (b-slot bar) 2)) (assert (= (c-slot bar) 3)))) ;;; some more CHANGE-CLASS testing, now that we have an ANSI-compliant ;;; version (thanks to Espen Johnsen) (defclass from-class () ((foo :initarg :foo :accessor foo))) (defclass to-class () ((foo :initarg :foo :accessor foo) (bar :initarg :bar :accessor bar))) (let* ((from (make-instance 'from-class :foo 1)) (to (change-class from 'to-class :bar 2))) (assert (= (foo to) 1)) (assert (= (bar to) 2))) ;;; Until Pierre Mai's patch (sbcl-devel 2002-06-18, merged in ;;; sbcl-0.7.4.39) the :MOST-SPECIFIC-LAST option had no effect. (defgeneric bug180 (x) (:method-combination list :most-specific-last)) (defmethod bug180 list ((x number)) 'number) (defmethod bug180 list ((x fixnum)) 'fixnum) (assert (equal (bug180 14) '(number fixnum))) ;;; printing a structure class should not loop indefinitely (or cause ;;; a stack overflow): (defclass test-printing-structure-class () ((slot :initarg :slot)) (:metaclass structure-class)) (print (make-instance 'test-printing-structure-class :slot 2)) ;;; structure-classes should behave nicely when subclassed (defclass super-structure () ((a :initarg :a :accessor a-accessor) (b :initform 2 :reader b-reader)) (:metaclass structure-class)) (defclass sub-structure (super-structure) ((c :initarg :c :writer c-writer :accessor c-accessor)) (:metaclass structure-class)) (let ((foo (make-instance 'sub-structure :a 1 :c 3))) (assert (= (a-accessor foo) 1)) (assert (= (b-reader foo) 2)) (assert (= (c-accessor foo) 3)) (setf (a-accessor foo) 4) (c-writer 5 foo) (assert (= (a-accessor foo) 4)) (assert (= (c-accessor foo) 5))) ;;; At least as of sbcl-0.7.4, PCL has code to support a special ;;; encoding of effective method functions for slot accessors as ;;; FIXNUMs. Given this special casing, it'd be easy for slot accessor ;;; functions to get broken in special ways even though ordinary ;;; generic functions work. As of sbcl-0.7.4 we didn't have any tests ;;; for that possibility. Now we have a few tests: (defclass fish () ((fin :reader ffin :writer ffin!) (tail :reader ftail :writer ftail!))) (defvar *fish* (make-instance 'fish)) (ffin! 'triangular-fin *fish*) (defclass cod (fish) ()) (defvar *cod* (make-instance 'cod)) (defparameter *clos-dispatch-side-fx* (make-array 0 :fill-pointer 0)) (defmethod ffin! (new-fin (cod cod)) (format t "~&about to set ~S fin to ~S~%" cod new-fin) (vector-push-extend '(cod) *clos-dispatch-side-fx*) (prog1 (call-next-method) (format t "~&done setting ~S fin to ~S~%" cod new-fin))) (defmethod ffin! :before (new-fin (cod cod)) (vector-push-extend '(:before cod) *clos-dispatch-side-fx*) (format t "~&exploring the CLOS dispatch zoo with COD fins~%")) (ffin! 'almost-triang-fin *cod*) (assert (eq (ffin *cod*) 'almost-triang-fin)) (assert (equalp #((:before cod) (cod)) *clos-dispatch-side-fx*)) ;;; Until sbcl-0.7.6.21, the long form of DEFINE-METHOD-COMBINATION ;;; ignored its options; Gerd Moellmann found and fixed the problem ;;; for cmucl (cmucl-imp 2002-06-18). (define-method-combination test-mc (x) ;; X above being a method-group-specifier ((primary () :required t)) `(call-method ,(first primary))) (defgeneric gf (obj) (:method-combination test-mc 1)) (defmethod gf (obj) obj) ;;; Until sbcl-0.7.7.20, some conditions weren't being signalled, and ;;; some others were of the wrong type: (macrolet ((assert-program-error (form) `(multiple-value-bind (value error) (ignore-errors ,form) (assert (null value)) (assert (typep error 'program-error))))) (assert-program-error (defclass foo001 () (a b a))) (assert-program-error (defclass foo002 () (a b) (:default-initargs x 'a x 'b))) (assert-program-error (defclass foo003 () ((a :allocation :class :allocation :class)))) (assert-program-error (defclass foo004 () ((a :silly t)))) ;; and some more, found by Wolfhard Buss and fixed for cmucl by Gerd ;; Moellmann in sbcl-0.7.8.x: (assert-program-error (progn (defmethod odd-key-args-checking (&key (key 42)) key) (odd-key-args-checking 3))) (assert (= (odd-key-args-checking) 42)) (assert (eq (odd-key-args-checking :key t) t)) ;; yet some more, fixed in sbcl-0.7.9.xx (assert-program-error (defclass foo005 () (:metaclass sb-pcl::funcallable-standard-class) (:metaclass 1))) (assert-program-error (defclass foo006 () ((a :reader (setf a))))) (assert-program-error (defclass foo007 () ((a :initarg 1)))) (assert-program-error (defclass foo008 () (a :initarg :a) (:default-initargs :a 1) (:default-initargs :a 2)))) ;;; DOCUMENTATION's argument-precedence-order wasn't being faithfully ;;; preserved through the bootstrap process until sbcl-0.7.8.39. ;;; (thanks to Gerd Moellmann) (let ((answer (documentation '+ 'function))) (assert (stringp answer)) (defmethod documentation ((x (eql '+)) y) "WRONG") (assert (string= (documentation '+ 'function) answer))) ;;; only certain declarations are permitted in DEFGENERIC (macrolet ((assert-program-error (form) `(multiple-value-bind (value error) (ignore-errors ,form) (assert (null value)) (assert (typep error 'program-error))))) (assert-program-error (defgeneric bogus-declaration (x) (declare (special y)))) (assert-program-error (defgeneric bogus-declaration2 (x) (declare (notinline concatenate))))) ;;; CALL-NEXT-METHOD should call NO-NEXT-METHOD if there is no next ;;; method. (defmethod no-next-method-test ((x integer)) (call-next-method)) (assert (null (ignore-errors (no-next-method-test 1)))) (defmethod no-next-method ((g (eql #'no-next-method-test)) m &rest args) 'success) (assert (eq (no-next-method-test 1) 'success)) (assert (null (ignore-errors (no-next-method-test 'foo)))) ;;; regression test for bug 176, following a fix that seems ;;; simultaneously to fix 140 while not exposing 176 (by Gerd ;;; Moellmann, merged in sbcl-0.7.9.12). (dotimes (i 10) (let ((lastname (intern (format nil "C176-~D" (1- i)))) (name (intern (format nil "C176-~D" i)))) (eval `(defclass ,name (,@(if (= i 0) nil (list lastname))) ())) (eval `(defmethod initialize-instance :after ((x ,name) &rest any) (declare (ignore any)))))) (defclass b176 () (aslot-176)) (defclass c176-0 (b176) ()) (assert (= 1 (setf (slot-value (make-instance 'c176-9) 'aslot-176) 1))) ;;; DEFINE-METHOD-COMBINATION was over-eager at checking for duplicate ;;; primary methods: (define-method-combination dmc-test-mc (&optional (order :most-specific-first)) ((around (:around)) (primary (dmc-test-mc) :order order :required t)) (let ((form (if (rest primary) `(and ,@(mapcar #'(lambda (method) `(call-method ,method)) primary)) `(call-method ,(first primary))))) (if around `(call-method ,(first around) (,@(rest around) (make-method ,form))) form))) (defgeneric dmc-test-mc (&key k) (:method-combination dmc-test-mc)) (defmethod dmc-test-mc dmc-test-mc (&key k) k) (dmc-test-mc :k 1) ;;; While I'm at it, DEFINE-METHOD-COMBINATION is defined to return ;;; the NAME argument, not some random method object. So: (assert (eq (define-method-combination dmc-test-return-foo) 'dmc-test-return-foo)) (assert (eq (define-method-combination dmc-test-return-bar :operator and) 'dmc-test-return-bar)) (assert (eq (define-method-combination dmc-test-return (&optional (order :most-specific-first)) ((around (:around)) (primary (dmc-test-return) :order order :required t)) (let ((form (if (rest primary) `(and ,@(mapcar #'(lambda (method) `(call-method ,method)) primary)) `(call-method ,(first primary))))) (if around `(call-method ,(first around) (,@(rest around) (make-method ,form))) form))) 'dmc-test-return)) ;;; DEFMETHOD should signal a PROGRAM-ERROR if an incompatible lambda ;;; list is given: (defmethod incompatible-ll-test-1 (x) x) (multiple-value-bind (result error) (ignore-errors (defmethod incompatible-ll-test-1 (x y) y)) (assert (null result)) (assert (typep error 'program-error))) (multiple-value-bind (result error) (ignore-errors (defmethod incompatible-ll-test-1 (x &rest y) y)) (assert (null result)) (assert (typep error 'program-error))) ;;; Sneakily using a bit of MOPness to check some consistency (assert (= (length (sb-pcl:generic-function-methods #'incompatible-ll-test-1)) 1)) (defmethod incompatible-ll-test-2 (x &key bar) bar) (multiple-value-bind (result error) (ignore-errors (defmethod incompatible-ll-test-2 (x) x)) (assert (null result)) (assert (typep error 'program-error))) (defmethod incompatible-ll-test-2 (x &rest y) y) (assert (= (length (sb-pcl:generic-function-methods #'incompatible-ll-test-2)) 1)) (defmethod incompatible-ll-test-2 ((x integer) &key bar) bar) (assert (= (length (sb-pcl:generic-function-methods #'incompatible-ll-test-2)) 2)) (assert (equal (incompatible-ll-test-2 t 1 2) '(1 2))) (assert (eq (incompatible-ll-test-2 1 :bar 'yes) 'yes)) ;;; Attempting to instantiate classes with forward references in their ;;; CPL should signal errors (FIXME: of what type?) (defclass never-finished-class (this-one-unfinished-too) ()) (multiple-value-bind (result error) (ignore-errors (make-instance 'never-finished-class)) (assert (null result)) (assert (typep error 'error))) (multiple-value-bind (result error) (ignore-errors (make-instance 'this-one-unfinished-too)) (assert (null result)) (assert (typep error 'error))) ;;; Classes with :ALLOCATION :CLASS slots should be subclassable (and ;;; weren't for a while in sbcl-0.7.9.xx) (defclass superclass-with-slot () ((a :allocation :class))) (defclass subclass-for-class-allocation (superclass-with-slot) ()) (make-instance 'subclass-for-class-allocation) ;;; bug #136: CALL-NEXT-METHOD was being a little too lexical, ;;; resulting in failure in the following: (defmethod call-next-method-lexical-args ((x integer)) x) (defmethod call-next-method-lexical-args :around ((x integer)) (let ((x (1+ x))) (call-next-method))) (assert (= (call-next-method-lexical-args 3) 3)) ;;; DEFINE-METHOD-COMBINATION with arguments was hopelessly broken ;;; until 0.7.9.5x (defvar *d-m-c-args-test* nil) (define-method-combination progn-with-lock () ((methods ())) (:arguments object) `(unwind-protect (progn (lock (object-lock ,object)) ,@(mapcar #'(lambda (method) `(call-method ,method)) methods)) (unlock (object-lock ,object)))) (defun object-lock (obj) (push "object-lock" *d-m-c-args-test*) obj) (defun unlock (obj) (push "unlock" *d-m-c-args-test*) obj) (defun lock (obj) (push "lock" *d-m-c-args-test*) obj) (defgeneric d-m-c-args-test (x) (:method-combination progn-with-lock)) (defmethod d-m-c-args-test ((x symbol)) (push "primary" *d-m-c-args-test*)) (defmethod d-m-c-args-test ((x number)) (error "foo")) (assert (equal (d-m-c-args-test t) '("primary" "lock" "object-lock"))) (assert (equal *d-m-c-args-test* '("unlock" "object-lock" "primary" "lock" "object-lock"))) (setf *d-m-c-args-test* nil) (ignore-errors (d-m-c-args-test 1)) (assert (equal *d-m-c-args-test* '("unlock" "object-lock" "lock" "object-lock"))) ;;; The walker (on which DEFMETHOD depended) didn't know how to handle ;;; SYMBOL-MACROLET properly. In fact, as of sbcl-0.7.10.20 it still ;;; doesn't, but it does well enough to compile the following without ;;; error (the problems remain in asking for a complete macroexpansion ;;; of an arbitrary form). (symbol-macrolet ((x 1)) (defmethod bug222 (z) (macrolet ((frob (form) `(progn ,form ,x))) (frob (print x))))) (assert (= (bug222 t) 1)) ;;; also, a test case to guard against bogus environment hacking: (eval-when (:compile-toplevel :load-toplevel :execute) (setq bug222-b 3)) ;;; this should at the least compile: (let ((bug222-b 1)) (defmethod bug222-b (z stream) (macrolet ((frob (form) `(progn ,form ,bug222-b))) (frob (format stream "~D~%" bug222-b))))) ;;; and it would be nice (though not specified by ANSI) if the answer ;;; were as follows: (let ((x (make-string-output-stream))) ;; not specified by ANSI (assert (= (bug222-b t x) 3)) ;; specified. (assert (char= (char (get-output-stream-string x) 0) #\1))) ;;; REINITIALIZE-INSTANCE, in the ctor optimization, wasn't checking ;;; for invalid initargs where it should: (defclass class234 () ()) (defclass subclass234 (class234) ()) (defvar *bug234* 0) (defun bug-234 () (reinitialize-instance (make-instance 'class234) :dummy 0)) (defun subbug-234 () (reinitialize-instance (make-instance 'subclass234) :dummy 0)) (assert (raises-error? (bug-234) program-error)) (defmethod shared-initialize :after ((i class234) slots &key dummy) (incf *bug234*)) (assert (typep (subbug-234) 'subclass234)) (assert (= *bug234* ;; once for MAKE-INSTANCE, once for REINITIALIZE-INSTANCE 2)) ;;; also, some combinations of MAKE-INSTANCE and subclassing missed ;;; new methods (Gerd Moellmann sbcl-devel 2002-12-29): (defclass class234-b1 () ()) (defclass class234-b2 (class234-b1) ()) (defvar *bug234-b* 0) (defun bug234-b () (make-instance 'class234-b2)) (compile 'bug234-b) (bug234-b) (assert (= *bug234-b* 0)) (defmethod initialize-instance :before ((x class234-b1) &rest args) (declare (ignore args)) (incf *bug234-b*)) (bug234-b) (assert (= *bug234-b* 1)) ;;; we should be able to make classes with uninterned names: (defclass #:class-with-uninterned-name () ()) ;;; SLOT-MISSING should be called when there are missing slots. (defclass class-with-all-slots-missing () ()) (defmethod slot-missing (class (o class-with-all-slots-missing) slot-name op &optional new-value) op) (assert (eq (slot-value (make-instance 'class-with-all-slots-missing) 'foo) 'slot-value)) (assert (eq (funcall (lambda (x) (slot-value x 'bar)) (make-instance 'class-with-all-slots-missing)) 'slot-value)) (assert (eq (funcall (lambda (x) (setf (slot-value x 'baz) 'baz)) (make-instance 'class-with-all-slots-missing)) 'setf)) ;;;; success (sb-ext:quit :unix-status 104)