;;;; tests for problems in the interface presented to the user/programmer ;;;; 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. (in-package :cl-user) (load "test-util.lisp") (load "compiler-test-util.lisp") (use-package :test-util) ;;;; properties of symbols, e.g. presence of doc strings for public symbols ;;; FIXME: It would probably be good to require here that every ;;; external symbol either has a doc string or has some good excuse ;;; (like being an accessor for a structure which has a doc string). ;;;; tests of interface machinery ;;; APROPOS should accept a package designator, not just a package, and ;;; furthermore do the right thing when it gets a package designator. ;;; (bug reported and fixed by Alexey Dejneka sbcl-devel 2001-10-17) (assert (< 0 (length (apropos-list "PRINT" :cl)) (length (apropos-list "PRINT")))) ;;; Further, it should correctly deal with the external-only flag (bug ;;; reported by cliini on #lisp IRC 2003-05-30, fixed in sbcl-0.8.0.1x ;;; by CSR) (assert (= (length (apropos-list "" "CL")) (length (apropos-list "" "CL" t)))) (assert (< 0 (length (apropos-list "" "SB-VM" t)) (length (apropos-list "" "SB-VM")))) ;;; DESCRIBE shouldn't fail on rank-0 arrays (bug reported and fixed ;;; by Lutz Euler sbcl-devel 2002-12-03) (describe #0a0) (describe #(1 2 3)) (describe #2a((1 2) (3 4))) ;;; TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and ;;; UPGRADED-COMPLEX-PART-TYPE should be able to deal with NIL as an ;;; environment argument (typep 1 'fixnum nil) (subtypep 'fixnum 'integer nil) (upgraded-array-element-type '(mod 5) nil) (upgraded-complex-part-type '(single-float 0.0 1.0) nil) ;;; We should have documentation for our extension package: (assert (documentation (find-package "SB-EXT") t)) ;;; DECLARE should not be a special operator (assert (not (special-operator-p 'declare))) ;;; WITH-TIMEOUT should accept more than one form in its body. (with-test (:name :with-timeout-forms) (handler-bind ((sb-ext:timeout #'continue)) (sb-ext:with-timeout 3 (sleep 2) (sleep 2)))) ;;; SLEEP should not cons except on 32-bit platforms when ;;; (> (mod seconds 1) (* most-positive-fixnum 1e-9)) (with-test (:name (sleep :non-consing) :fails-on :win32) (ctu:assert-no-consing (sleep 0.00001s0)) (locally (declare (notinline sleep)) (ctu:assert-no-consing (sleep 0.00001s0)) (ctu:assert-no-consing (sleep 0.00001d0)) (ctu:assert-no-consing (sleep 1/100000003)))) ;;; Changes to make SLEEP cons less led to SLEEP ;;; not sleeping at all on 32-bit platforms when ;;; (> (mod seconds 1) (* most-positive-fixnum 1e-9)). (with-test (:name :bug-1194673) (assert (eq :timeout (handler-case (with-timeout 0.01 (sleep 0.6)) (timeout () :timeout))))) ;;; SLEEP should work with large integers as well (with-test (:name (sleep pretty-much-forever)) (assert (eq :timeout (handler-case (sb-ext:with-timeout 1 (sleep (ash 1 (* 2 sb-vm:n-word-bits)))) (sb-ext:timeout () :timeout))))) ;;; DOCUMENTATION should return nil, not signal slot-unbound (documentation 'fixnum 'type) (documentation 'class 'type) (documentation (find-class 'class) 'type) (documentation 'foo 'structure) ;;; DECODE-UNIVERSAL-TIME should accept second-resolution time-zones. (macrolet ((test (ut time-zone list) (destructuring-bind (sec min hr date mon yr day tz) list `(multiple-value-bind (sec min hr date mon yr day dst tz) (decode-universal-time ,ut ,time-zone) (declare (ignore dst)) (assert (= sec ,sec)) (assert (= min ,min)) (assert (= hr ,hr)) (assert (= date ,date)) (assert (= mon ,mon)) (assert (= yr ,yr)) (assert (= day ,day)) (assert (= tz ,tz)))))) (test (* 86400 365) -1/3600 (1 0 0 1 1 1901 1 -1/3600)) (test (* 86400 365) 0 (0 0 0 1 1 1901 1 0)) (test (* 86400 365) 1/3600 (59 59 23 31 12 1900 0 1/3600))) ;;; DECODE-UNIVERSAL-TIME shouldn't fail when the time is outside UNIX ;;; 32-bit time_t and a timezone wasn't passed (decode-universal-time 0 nil) ;;; ENCODE-UNIVERSAL-TIME should be able to encode the universal time ;;; 0 when passed a representation in a timezone where the ;;; representation of 0 as a decoded time is in 1899. (encode-universal-time 0 0 23 31 12 1899 1) ;;; DISASSEMBLE shouldn't fail on purified functions (disassemble 'cl:+) (disassemble 'sb-ext:run-program) ;;; minimal test of GC: see stress-gc.{sh,lisp} for a more ;;; comprehensive test. (loop repeat 2 do (compile nil '(lambda (x) x)) do (sb-ext:gc :full t)) ;;; On x86-64, the instruction definitions for CMP*[PS][SD] were broken ;;; so that the disassembler threw an error when they were used with ;;; one operand in memory. (with-test (:name :bug-814702) (disassemble (lambda (x) (= #C(2.0f0 3.0f0) (the (complex single-float) x)))) (disassemble (lambda (x y) (= (the (complex single-float) x) (the (complex single-float) y))))) ;;; Check that SLEEP called with ratios (with no common factors with ;;; 1000000000, and smaller than 1/1000000000) works more or less as ;;; expected. (with-test (:name :sleep-ratios) (let ((fun0a (compile nil '(lambda () (sleep 1/7)))) (fun0b (compile nil '(lambda () (sleep 1/100000000000000000000000000)))) (fun1 (compile nil '(lambda (x) (sleep x)))) (start-time (get-universal-time))) (sleep 1/7) (sleep 1/100000000000000000000000000) (funcall fun0a) (funcall fun0b) (funcall fun1 1/7) (funcall fun1 1/100000000000000000000000000) (assert (< (- (get-universal-time) start-time) 2))))