;;;; arithmetic tests with no side effects ;;;; 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. (cl:in-package :cl-user) ;;; Once upon a time, in the process of porting CMUCL's SPARC backend ;;; to SBCL, multiplications were excitingly broken. While it's ;;; unlikely that anything with such fundamental arithmetic errors as ;;; these are going to get this far, it's probably worth checking. (macrolet ((test (op res1 res2) `(progn (assert (= (,op 4 2) ,res1)) (assert (= (,op 2 4) ,res2)) (assert (= (funcall (compile nil '(lambda (x y) (,op x y))) 4 2) ,res1)) (assert (= (funcall (compile nil '(lambda (x y) (,op x y))) 2 4) ,res2))))) (test + 6 6) (test - 2 -2) (test * 8 8) (test / 2 1/2) (test expt 16 16)) ;;; In a bug reported by Wolfhard Buss on cmucl-imp 2002-06-18 (BUG ;;; 184), sbcl didn't catch all divisions by zero, notably divisions ;;; of bignums and ratios by 0. Fixed in sbcl-0.7.6.13. (assert (raises-error? (/ 2/3 0) division-by-zero)) (assert (raises-error? (/ (1+ most-positive-fixnum) 0) division-by-zero)) ;;; In a bug reported by Raymond Toy on cmucl-imp 2002-07-18, (COERCE ;;; '(COMPLEX FLOAT)) was failing to return a complex ;;; float; a patch was given by Wolfhard Buss cmucl-imp 2002-07-19. (assert (= (coerce 1 '(complex float)) #c(1.0 0.0))) (assert (= (coerce 1/2 '(complex float)) #c(0.5 0.0))) (assert (= (coerce 1.0d0 '(complex float)) #c(1.0d0 0.0d0))) ;;; (COERCE #c( ) '(complex float)) resulted in ;;; an error up to 0.8.17.31 (assert (= (coerce #c(1 2) '(complex float)) #c(1.0 2.0))) ;;; COERCE also sometimes failed to verify that a particular coercion ;;; was possible (in particular coercing rationals to bounded float ;;; types. (assert (raises-error? (coerce 1 '(float 2.0 3.0)) type-error)) (assert (raises-error? (coerce 1 '(single-float -1.0 0.0)) type-error)) (assert (eql (coerce 1 '(single-float -1.0 2.0)) 1.0)) ;;; ANSI says MIN and MAX should signal TYPE-ERROR if any argument ;;; isn't REAL. SBCL 0.7.7 didn't in the 1-arg case. (reported as a ;;; bug in CMU CL on #lisp IRC by lrasinen 2002-09-01) (assert (null (ignore-errors (min '(1 2 3))))) (assert (= (min -1) -1)) (assert (null (ignore-errors (min 1 #(1 2 3))))) (assert (= (min 10 11) 10)) (assert (null (ignore-errors (min (find-package "CL") -5.0)))) (assert (= (min 5.0 -3) -3)) (assert (null (ignore-errors (max #c(4 3))))) (assert (= (max 0) 0)) (assert (null (ignore-errors (max "MIX" 3)))) (assert (= (max -1 10.0) 10.0)) (assert (null (ignore-errors (max 3 #'max)))) (assert (= (max -3 0) 0)) ;;; (CEILING x 2^k) was optimized incorrectly (loop for divisor in '(-4 4) for ceiler = (compile nil `(lambda (x) (declare (fixnum x)) (declare (optimize (speed 3))) (ceiling x ,divisor))) do (loop for i from -5 to 5 for exact-q = (/ i divisor) do (multiple-value-bind (q r) (funcall ceiler i) (assert (= (+ (* q divisor) r) i)) (assert (<= exact-q q)) (assert (< q (1+ exact-q)))))) ;;; (TRUNCATE x 2^k) was optimized incorrectly (loop for divisor in '(-4 4) for truncater = (compile nil `(lambda (x) (declare (fixnum x)) (declare (optimize (speed 3))) (truncate x ,divisor))) do (loop for i from -9 to 9 for exact-q = (/ i divisor) do (multiple-value-bind (q r) (funcall truncater i) (assert (= (+ (* q divisor) r) i)) (assert (<= (abs q) (abs exact-q))) (assert (< (abs exact-q) (1+ (abs q))))))) ;;; CEILING had a corner case, spotted by Paul Dietz (assert (= (ceiling most-negative-fixnum (1+ most-positive-fixnum)) -1)) ;;; give any optimizers of constant multiplication a light testing. ;;; 100 may seem low, but (a) it caught CSR's initial errors, and (b) ;;; before checking in, CSR tested with 10000. So one hundred ;;; checkins later, we'll have doubled the coverage. (dotimes (i 100) (let* ((x (random most-positive-fixnum)) (x2 (* x 2)) (x3 (* x 3))) (let ((fn (handler-bind ((sb-ext:compiler-note (lambda (c) (when (<= x3 most-positive-fixnum) (error c))))) (compile nil `(lambda (y) (declare (optimize speed) (type (integer 0 3) y)) (* y ,x)))))) (unless (and (= (funcall fn 0) 0) (= (funcall fn 1) x) (= (funcall fn 2) x2) (= (funcall fn 3) x3)) (error "bad results for ~D" x))))) ;;; Bugs reported by Paul Dietz: ;;; (GCD 0 x) must return (abs x) (dolist (x (list -10 (* 3 most-negative-fixnum))) (assert (= (gcd 0 x) (abs x)))) ;;; LCM returns a non-negative number (assert (= (lcm 4 -10) 20)) (assert (= (lcm 0 0) 0)) ;;; PPC bignum arithmetic bug: (multiple-value-bind (quo rem) (truncate 291351647815394962053040658028983955 10000000000000000000000000) (assert (= quo 29135164781)) (assert (= rem 5394962053040658028983955))) ;;; x86 LEA bug: (assert (= (funcall (compile nil '(lambda (x) (declare (bit x)) (+ x #xf0000000))) 1) #xf0000001)) ;;; LOGBITP on bignums: (dolist (x '(((1+ most-positive-fixnum) 1 nil) ((1+ most-positive-fixnum) -1 t) ((1+ most-positive-fixnum) (1+ most-positive-fixnum) nil) ((1+ most-positive-fixnum) (1- most-negative-fixnum) t) (1 (ash most-negative-fixnum 1) nil) (#.(- sb-vm:n-word-bits sb-vm:n-fixnum-tag-bits 1) most-negative-fixnum t) (#.(1+ (- sb-vm:n-word-bits sb-vm:n-fixnum-tag-bits 1)) (ash most-negative-fixnum 1) t) (#.(+ 2 (- sb-vm:n-word-bits sb-vm:n-fixnum-tag-bits 1)) (ash most-negative-fixnum 1) t) (#.(+ sb-vm:n-word-bits 32) (ash most-negative-fixnum #.(+ 32 sb-vm:n-fixnum-tag-bits 2)) nil) (#.(+ sb-vm:n-word-bits 33) (ash most-negative-fixnum #.(+ 32 sb-vm:n-fixnum-tag-bits 2)) t))) (destructuring-bind (index int result) x (assert (eq (eval `(logbitp ,index ,int)) result)))) ;;; off-by-1 type inference error for %DPB and %DEPOSIT-FIELD: (let ((f (compile nil '(lambda (b) (integer-length (dpb b (byte 4 28) -1005)))))) (assert (= (funcall f 1230070) 32))) (let ((f (compile nil '(lambda (b) (integer-length (deposit-field b (byte 4 28) -1005)))))) (assert (= (funcall f 1230070) 32))) ;;; type inference leading to an internal compiler error: (let ((f (compile nil '(lambda (x) (declare (type fixnum x)) (ldb (byte 0 0) x))))) (assert (= (funcall f 1) 0)) (assert (= (funcall f most-positive-fixnum) 0)) (assert (= (funcall f -1) 0))) ;;; Alpha bignum arithmetic bug: (assert (= (* 966082078641 419216044685) 404997107848943140073085)) ;;; Alpha smallnum arithmetic bug: (assert (= (ash -129876 -1026) -1)) ;;; Alpha middlenum (yes, really! Affecting numbers between 2^32 and ;;; 2^64 :) arithmetic bug (let ((fn (compile nil '(LAMBDA (A B C D) (DECLARE (TYPE (INTEGER -1621 -513) A) (TYPE (INTEGER -3 34163) B) (TYPE (INTEGER -9485132993 81272960) C) (TYPE (INTEGER -255340814 519943) D) (IGNORABLE A B C D) (OPTIMIZE (SPEED 3) (SAFETY 1) (DEBUG 1))) (TRUNCATE C (MIN -100 4149605)))))) (assert (= (funcall fn -1332 5864 -6963328729 -43789079) 69633287))) ;;; Here's another fantastic Alpha backend bug: the code to load ;;; immediate 64-bit constants into a register was wrong. (let ((fn (compile nil '(LAMBDA (A B C D) (DECLARE (TYPE (INTEGER -3563 2733564) A) (TYPE (INTEGER -548947 7159) B) (TYPE (INTEGER -19 0) C) (TYPE (INTEGER -2546009 0) D) (IGNORABLE A B C D) (OPTIMIZE (SPEED 3) (SAFETY 1) (DEBUG 1))) (CASE A ((89 125 16) (ASH A (MIN 18 -706))) (T (DPB -3 (BYTE 30 30) -1))))))) (assert (= (funcall fn 1227072 -529823 -18 -792831) -2147483649))) ;;; ASH of a negative bignum by a bignum count would erroneously ;;; return 0 prior to sbcl-0.8.4.4 (assert (= (ash (1- most-negative-fixnum) (1- most-negative-fixnum)) -1)) ;;; Whoops. Too much optimization in division operators for 0 ;;; divisor. (macrolet ((frob (name) `(let ((fn (compile nil '(lambda (x) (declare (optimize speed) (fixnum x)) (,name x 0))))) (assert (raises-error? (funcall fn 1) division-by-zero))))) (frob mod) (frob truncate) (frob rem) (frob /) (frob floor) (frob ceiling)) ;; Check that the logic in SB-KERNEL::BASIC-COMPARE for doing fixnum/float ;; comparisons without rationalizing the floats still gives the right anwers ;; in the edge cases (had a fencepost error). (macrolet ((test (range type sign) `(let (ints floats (start (- ,(find-symbol (format nil "MOST-~A-EXACTLY-~A-FIXNUM" sign type) :sb-kernel) ,range))) (dotimes (i (1+ (* ,range 2))) (let* ((x (+ start i)) (y (coerce x ',type))) (push x ints) (push y floats))) (dolist (i ints) (dolist (f floats) (dolist (op '(< <= = >= >)) (unless (eq (funcall op i f) (funcall op i (rationalize f))) (error "(not (eq (~a ~a ~f) (~a ~a ~a)))~%" op i f op i (rationalize f))) (unless (eq (funcall op f i) (funcall op (rationalize f) i)) (error "(not (eq (~a ~f ~a) (~a ~a ~a)))~%" op f i op (rationalize f) i)))))))) (test 32 double-float negative) (test 32 double-float positive) (test 32 single-float negative) (test 32 single-float positive)) ;; x86-64 sign-extension bug found using pfdietz's random tester. (assert (= 286142502 (funcall (lambda () (declare (notinline logxor)) (min (logxor 0 0 0 286142502)))))) ;; Small bugs in LOGCOUNT can still allow SBCL to be built and thus go ;; unnoticed, so check more thoroughly here. (with-test (:name :logcount) (flet ((test (x n) (unless (= (logcount x) n) (error "logcount failure for ~a" x)))) ;; Test with some patterns with well known number of ones/zeroes ... (dotimes (i 128) (let ((x (ash 1 i))) (test x 1) (test (- x) i) (test (1- x) i))) ;; ... and with some random integers of varying length. (flet ((test-logcount (x) (declare (type integer x)) (do ((result 0 (1+ result)) (x (if (minusp x) (lognot x) x) (logand x (1- x)))) ((zerop x) result)))) (dotimes (i 200) (let ((x (random (ash 1 i)))) (test x (test-logcount x)) (test (- x) (test-logcount (- x)))))))) ;; 1.0 had a broken ATANH on win32 (with-test (:name :atanh) (assert (= (atanh 0.9d0) 1.4722194895832204d0))) ;; Test some cases of integer operations with constant arguments (with-test (:name :constant-integers) (labels ((test-forms (op x y header &rest forms) (let ((val (funcall op x y))) (dolist (form forms) (let ((new-val (funcall (compile nil (append header form)) x y))) (unless (eql val new-val) (error "~S /= ~S: ~S ~S ~S~%" val new-val (append header form) x y)))))) (test-case (op x y type) (test-forms op x y `(lambda (x y &aux z) (declare (type ,type x y) (ignorable x y z) (notinline identity) (optimize speed (safety 0)))) `((,op x ,y)) `((setf z (,op x ,y)) (identity x) z) `((values (,op x ,y) x)) `((,op ,x y)) `((setf z (,op ,x y)) (identity y) z) `((values (,op ,x y) y)) `((identity x) (,op x ,y)) `((identity x) (setf z (,op x ,y)) (identity x) z) `((identity x) (values (,op x ,y) x)) `((identity y) (,op ,x y)) `((identity y) (setf z (,op ,x y)) (identity y) z) `((identity y) (values (,op ,x y) y)))) (test-op (op) (let ((ub `(unsigned-byte ,sb-vm:n-word-bits)) (sb `(signed-byte ,sb-vm:n-word-bits))) (loop for (x y type) in `((2 1 fixnum) (2 1 ,ub) (2 1 ,sb) (,(1+ (ash 1 28)) ,(1- (ash 1 28)) fixnum) (,(+ 3 (ash 1 30)) ,(+ 2 (ash 1 30)) ,ub) (,(- -2 (ash 1 29)) ,(- 3 (ash 1 29)) ,sb) ,@(when (> sb-vm:n-word-bits 32) `((,(1+ (ash 1 29)) ,(1- (ash 1 29)) fixnum) (,(1+ (ash 1 31)) ,(1- (ash 1 31)) ,ub) (,(- -2 (ash 1 31)) ,(- 3 (ash 1 30)) ,sb) (,(ash 1 40) ,(ash 1 39) fixnum) (,(ash 1 40) ,(ash 1 39) ,ub) (,(ash 1 40) ,(ash 1 39) ,sb))) ;; fixnums that can be represented as 32-bit ;; sign-extended immediates on x86-64 ,@(when (and (> sb-vm:n-word-bits 32) (< sb-vm:n-fixnum-tag-bits 3)) `((,(1+ (ash 1 (- 31 sb-vm:n-fixnum-tag-bits))) ,(1- (ash 1 (- 32 sb-vm:n-fixnum-tag-bits))) fixnum)))) do (test-case op x y type) (test-case op x x type))))) (mapc #'test-op '(+ - * truncate < <= = >= > eql eq)))) ;; GCD used to sometimes return negative values. The following did, on 32 bit ;; builds. (with-test (:name :gcd) ;; from lp#413680 (assert (plusp (gcd 20286123923750474264166990598656 680564733841876926926749214863536422912))) ;; from lp#516750 (assert (plusp (gcd 2596102012663483082521318626691873 2596148429267413814265248164610048)))) (with-test (:name :expt-zero-zero) ;; Check that (expt 0.0 0.0) and (expt 0 0.0) signal error, but (expt 0.0 0) ;; returns 1.0 (assert (raises-error? (expt 0.0 0.0) sb-int:arguments-out-of-domain-error)) (assert (raises-error? (expt 0 0.0) sb-int:arguments-out-of-domain-error)) (assert (eql (expt 0.0 0) 1.0))) (with-test (:name :multiple-constant-folding) (let ((*random-state* (make-random-state t))) (flet ((make-args () (let (args vars) (loop repeat (1+ (random 12)) do (if (zerop (random 2)) (let ((var (gensym))) (push var args) (push var vars)) (push (- (random 21) 10) args))) (values args vars)))) (dolist (op '(+ * logior logxor logand logeqv gcd lcm - /)) (loop repeat 10 do (multiple-value-bind (args vars) (make-args) (let ((fast (compile nil `(lambda ,vars (,op ,@args)))) (slow (compile nil `(lambda ,vars (declare (notinline ,op)) (,op ,@args))))) (loop repeat 3 do (let* ((call-args (loop repeat (length vars) collect (- (random 21) 10))) (fast-result (handler-case (apply fast call-args) (division-by-zero () :div0))) (slow-result (handler-case (apply slow call-args) (division-by-zero () :div0)))) (if (eql fast-result slow-result) (print (list :ok `(,op ,@args) :=> fast-result)) (error "oops: ~S, ~S" args call-args))))))))))) ;;; (TRUNCATE ) is optimized ;;; to use multiplication instead of division. This propagates to FLOOR, ;;; MOD and REM. Test that the transform is indeed triggered and test ;;; several cases for correct results. (with-test (:name (:integer-division-using-multiplication :used) :skipped-on '(not (or :x86-64 :x86))) (dolist (fun '(truncate floor ceiling mod rem)) (let* ((foo (compile nil `(lambda (x) (declare (optimize (speed 3) (space 1) (compilation-speed 0)) (type (unsigned-byte ,sb-vm:n-word-bits) x)) (,fun x 9)))) (disassembly (with-output-to-string (s) (disassemble foo :stream s)))) ;; KLUDGE copied from test :float-division-using-exact-reciprocal ;; in compiler.pure.lisp. (assert (and (not (search "DIV" disassembly)) (search "MUL" disassembly)))))) (with-test (:name (:integer-division-using-multiplication :correctness)) (let ((*random-state* (make-random-state t))) (dolist (dividend-type `((unsigned-byte ,sb-vm:n-word-bits) (and fixnum unsigned-byte) (integer 10000 10100))) (dolist (divisor `(;; Some special cases from the paper 7 10 14 641 274177 ;; Range extremes 3 ,most-positive-fixnum ,(1- (expt 2 sb-vm:n-word-bits)) ;; Some random values ,@(loop for i from 8 to sb-vm:n-word-bits for r = (random (expt 2 i)) ;; We don't want 0, 1 and powers of 2. when (not (zerop (logand r (1- r)))) collect r))) (dolist (fun '(truncate ceiling floor mod rem)) (let ((foo (compile nil `(lambda (x) (declare (optimize (speed 3) (space 1) (compilation-speed 0)) (type ,dividend-type x)) (,fun x ,divisor))))) (dolist (dividend `(0 1 ,most-positive-fixnum ,(1- divisor) ,divisor ,(1- (* divisor 2)) ,(* divisor 2) ,@(loop repeat 4 collect (+ 10000 (random 101))) ,@(loop for i from 4 to sb-vm:n-word-bits for pow = (expt 2 (1- i)) for r = (+ pow (random pow)) collect r))) (when (typep dividend dividend-type) (multiple-value-bind (q1 r1) (funcall foo dividend) (multiple-value-bind (q2 r2) (funcall fun dividend divisor) (unless (and (= q1 q2) (eql r1 r2)) (error "bad results for ~s with dividend type ~s" (list fun dividend divisor) dividend-type)))))))))))) ;; The fast path for logbitp underestimated sb!vm:n-positive-fixnum-bits ;; for > 61 bit fixnums. (with-test (:name :logbitp-wide-fixnum) (assert (not (logbitp (1- (integer-length most-positive-fixnum)) most-negative-fixnum)))) ;; EXPT dispatches in a complicated way on the types of its arguments. ;; Check that all possible combinations are covered. (with-test (:name (:expt :argument-type-combinations)) (let ((numbers '(2 ; fixnum 3/5 ; ratio 1.2f0 ; single-float 2.0d0 ; double-float #c(3/5 1/7) ; complex rational #c(1.2f0 1.3f0) ; complex single-float #c(2.0d0 3.0d0))) ; complex double-float (bignum (expt 2 64)) results) (dolist (base (cons bignum numbers)) (dolist (power numbers) (format t "(expt ~s ~s) => " base power) (let ((result (expt base power))) (format t "~s~%" result) (push result results)))) (assert (every #'numberp results)))) (with-test (:name :bug-741564) ;; The bug was that in (expt <(complex double-float)>) the ;; calculation was partially done only to single-float precision, ;; making the complex double-float result too unprecise. Some other ;; combinations of argument types were affected, too; test that all ;; of them are good to double-float precision. (labels ((nearly-equal-p (x y) "Are the arguments equal to nearly double-float precision?" (declare (type double-float x y)) (< (/ (abs (- x y)) (abs y)) (* double-float-epsilon 4))) ; Differences in the two least ; significant mantissa bits ; are OK. (test-complex (x y) (and (nearly-equal-p (realpart x) (realpart y)) (nearly-equal-p (imagpart x) (imagpart y)))) (print-result (msg base power got expected) (format t "~a (expt ~s ~s)~%got ~s~%expected ~s~%" msg base power got expected))) (let ((n-broken 0)) (flet ((test (base power coerce-to-type) (let* ((got (expt base power)) (expected (expt (coerce base coerce-to-type) power)) (result (test-complex got expected))) (print-result (if result "Good:" "Bad:") base power got expected) (unless result (incf n-broken))))) (dolist (base (list 2 ; fixnum (expt 2 64) ; bignum 3/5 ; ratio 2.0f0)) ; single-float (let ((power #c(-2.5d0 -4.5d0))) ; complex double-float (test base power 'double-float))) (dolist (base (list #c(2.0f0 3.0f0) ; complex single-float #c(2 3) ; complex fixnum (complex (expt 2 64) (expt 2 65)) ; complex bignum #c(3/5 1/7))) ; complex ratio (dolist (power (list #c(-2.5d0 -4.5d0) ; complex double-float -2.5d0)) ; double-float (test base power '(complex double-float))))) (when (> n-broken 0) (error "Number of broken combinations: ~a" n-broken))))) (with-test (:name (:ldb :rlwinm :ppc)) (let ((one (compile nil '(lambda (a) (ldb (byte 9 27) a)))) (two (compile nil '(lambda (a) (declare (type (integer -3 57216651) a)) (ldb (byte 9 27) a))))) (assert (= 0 (- (funcall one 10) (funcall two 10)))))) ;; The ISQRT implementation is sufficiently complicated that it should ;; be tested. (with-test (:name :isqrt) (labels ((test (x) (let* ((r (isqrt x)) (r2 (expt r 2)) (s2 (expt (1+ r) 2))) (unless (and (<= r2 x) (> s2 x)) (error "isqrt failure for ~a" x)))) (tests (x) (test x) (let ((x2 (expt x 2))) (test x2) (test (1+ x2)) (test (1- x2))))) (test most-positive-fixnum) (test (1+ most-positive-fixnum)) (loop for i from 1 to 200 for pow = (expt 2 (1- i)) for j = (+ pow (random pow)) do (tests i) (tests j)) (dotimes (i 10) (tests (random (expt 2 (+ 1000 (random 10000)))))))) ;; bug 1026634 (reported by Eric Marsden on sbcl-devel) (with-test (:name :recursive-cut-to-width) (assert (eql (funcall (compile nil `(lambda (x) (declare (optimize (space 3)) (type (integer 12417236377505266230 12417274239874990070) x)) (logand 8459622733968096971 x))) 12417237222845306758) 2612793697039849090))) ;; Also reported by Eric Marsden on sbcl-devel (2013-06-06) (with-test (:name :more-recursive-cut-to-width) (assert (eql (funcall (compile nil `(lambda (a b) (declare (optimize (speed 2) (safety 0))) (logand (the (eql 16779072918521075607) a) (the (integer 21371810342718833225 21371810343571293860) b)))) 16779072918521075607 21371810342718833263) 2923729245085762055))) (with-test (:name :complicated-logand-identity) (loop for k from -8 upto 8 do (loop for min from -16 upto 16 do (loop for max from min upto 16 do (let ((f (compile nil `(lambda (x) (declare (type (integer ,min ,max) x)) (logand x ,k))))) (loop for x from min upto max do (assert (eql (logand x k) (funcall f x))))))))) (with-test (:name :complicated-logior-identity) (loop for k from -8 upto 8 do (loop for min from -16 upto 16 do (loop for max from min upto 16 do (let ((f (compile nil `(lambda (x) (declare (type (integer ,min ,max) x)) (logior x ,k))))) (loop for x from min upto max do (assert (eql (logior x k) (funcall f x))))))))) (with-test (:name :ldb-negative-index-no-error) (assert (raises-error? (funcall (compile nil `(lambda (x y) (ldb (byte x y) 100))) -1 -2))) (assert (raises-error? (funcall (compile nil `(lambda (x y) (mask-field (byte x y) 100))) -1 -2))) (assert (raises-error? (funcall (compile nil `(lambda (x y) (dpb 0 (byte x y) 100))) -1 -2))) (assert (raises-error? (funcall (compile nil `(lambda (x y) (deposit-field 0 (byte x y) 100))) -1 -2))))