3 # From Tom Phoenix <rootbeer@teleport.com> 22 Feb 1997
4 # Based upon a test script by kgb@ast.cam.ac.uk (Karl Glazebrook)
6 # Looking for the hints? You're in the right place.
7 # The hints are near each test, so search for "TEST #", where
8 # the pound sign is replaced by the number of the test.
10 # I'd like to include some more robust tests, but anything
11 # too subtle to be detected here would require a time-consuming
12 # test. Also, of course, we're here to detect only flaws in Perl;
13 # if there are flaws in the underlying system rand, that's not
14 # our responsibility. But if you want better tests, see
15 # The Art of Computer Programming, Donald E. Knuth, volume 2,
16 # chapter 3. ISBN 0-201-03822-6 (v. 2)
28 srand; # Shouldn't need this with 5.004...
29 # But I'll include it now and test for
30 # whether we needed it later.
32 my $reps = 1000; # How many times to try rand each time.
33 # May be changed, but should be over 500.
34 # The more the better! (But slower.)
37 # Takes a small integer and returns the number of one-bits in it.
39 my $bits = sprintf "%o", $_[0];
40 while (length $bits) {
41 $total += (0,1,1,2,1,2,2,3)[chop $bits]; # Oct to bits
46 # First, let's see whether randbits is set right
48 my($max, $min, $sum); # Characteristics of rand
49 my($off, $shouldbe); # Problems with randbits
50 my($dev, $bits); # Number of one bits
51 my $randbits = $Config{randbits};
52 $max = $min = rand(1);
55 if ($n < 0.0 or $n >= 1.0) {
57 # WHOA THERE! \$Config{drand01} is set to '$Config{drand01}',
58 # but that apparently produces values < 0.0 or >= 1.0.
59 # Make sure \$Config{drand01} is a valid expression in the
60 # C-language, and produces values in the range [0.0,1.0).
67 $bits += bits($n * 256); # Don't be greedy; 8 is enough
68 # It's too many if randbits is less than 8!
69 # But that should never be the case... I hope.
70 # Note: If you change this, you must adapt the
71 # formula for absolute standard deviation, below.
72 $max = $n if $n > $max;
73 $min = $n if $n < $min;
79 # This test checks for one of Perl's most frequent
80 # mis-configurations. Your system's documentation
81 # for rand(2) should tell you what value you need
82 # for randbits. Usually the diagnostic message
83 # has the right value as well. Just fix it and
84 # recompile, and you'll usually be fine. (The main
85 # reason that the diagnostic message might get the
86 # wrong value is that Config.pm is incorrect.)
88 if ($max <= 0 or $max >= (2 ** $randbits)) {# Just in case...
89 print "# max=[$max] min=[$min]\nnot ok 1\n";
90 print "# This perl was compiled with randbits=$randbits\n";
91 print "# which is _way_ off. Or maybe your system rand is broken,\n";
92 print "# or your C compiler can't multiply, or maybe Martians\n";
93 print "# have taken over your computer. For starters, see about\n";
94 print "# trying a better value for randbits, probably smaller.\n";
95 # If that isn't the problem, we'll have
96 # to put d_martians into Config.pm
97 print "# Skipping remaining tests until randbits is fixed.\n";
101 $off = log($max) / log(2); # log2
102 $off = int($off) + ($off > 0); # Next more positive int
104 $shouldbe = $Config{randbits} + $off;
105 print "# max=[$max] min=[$min]\nnot ok 1\n";
106 print "# This perl was compiled with randbits=$randbits on $^O.\n";
107 print "# Consider using randbits=$shouldbe instead.\n";
108 # And skip the remaining tests; they would be pointless now.
109 print "# Skipping remaining tests until randbits is fixed.\n";
117 # This should always be true: 0 <= rand(1) < 1
118 # If this test is failing, something is seriously wrong,
119 # either in perl or your system's rand function.
121 if ($min < 0 or $max >= 1) { # Slightly redundant...
123 print "# min too low\n" if $min < 0;
124 print "# max too high\n" if $max >= 1;
131 # This is just a crude test. The average number produced
132 # by rand should be about one-half. But once in a while
133 # it will be relatively far away. Note: This test will
134 # occasionally fail on a perfectly good system!
135 # See the hints for test 4 to see why.
138 if ($sum < 0.4 or $sum > 0.6) {
139 print "not ok 3\n# Average random number is far from 0.5\n";
146 # NOTE NOTE NOTE NOTE NOTE NOTE NOTE NOTE NOTE NOTE
147 # This test will fail .1% of the time on a normal system.
149 # This test asks you to see these hints 100% of the time!
150 # NOTE NOTE NOTE NOTE NOTE NOTE NOTE NOTE NOTE NOTE
152 # There is probably no reason to be alarmed that
153 # something is wrong with your rand function. But,
154 # if you're curious or if you can't help being
155 # alarmed, keep reading.
157 # This is a less-crude test than test 3. But it has
158 # the same basic flaw: Unusually distributed random
159 # values should occasionally appear in every good
160 # random number sequence. (If you flip a fair coin
161 # twenty times every day, you'll see it land all
162 # heads about one time in a million days, on the
163 # average. That might alarm you if you saw it happen
166 # So, if this test failed on you once, run it a dozen
167 # times. If it keeps failing, it's likely that your
168 # rand is bogus. If it keeps passing, it's likely
169 # that the one failure was bogus. If it's a mix,
170 # read on to see about how to interpret the tests.
172 # The number printed in square brackets is the
173 # standard deviation, a statistical measure
174 # of how unusual rand's behavior seemed. It should
175 # fall in these ranges with these *approximate*
178 # under 1 68.26% of the time
179 # 1-2 27.18% of the time
180 # 2-3 4.30% of the time
181 # over 3 0.26% of the time
183 # If the numbers you see are not scattered approximately
184 # (not exactly!) like that table, check with your vendor
185 # to find out what's wrong with your rand. Or with this
188 # Calculating absoulute standard deviation for number of bits set
189 # (eight bits per rep)
190 $dev = abs ($bits - $reps * 4) / sqrt($reps * 2);
193 print "ok 4\n"; # 95% of the time.
194 print "# Your rand seems fine. If this test failed\n";
195 print "# previously, you may want to run it again.\n";
196 } elsif ($dev < 2.575) {
197 print "ok 4\n# In here about 4% of the time. Hmmm...\n";
198 print "# This is ok, but suspicious. But it will happen\n";
199 print "# one time out of 25, more or less.\n";
200 print "# You should run this test again to be sure.\n";
201 } elsif ($dev < 3.3) {
202 print "ok 4\n# In this range about 1% of the time.\n";
203 print "# This is very suspicious. It will happen only\n";
204 print "# about one time out of 100, more or less.\n";
205 print "# You should run this test again to be sure.\n";
206 } elsif ($dev < 3.9) {
207 print "not ok 4\n# In this range very rarely.\n";
208 print "# This is VERY suspicious. It will happen only\n";
209 print "# about one time out of 1000, more or less.\n";
210 print "# You should run this test again to be sure.\n";
212 print "not ok 4\n# Seriously whacked.\n";
213 print "# This is VERY VERY suspicious.\n";
214 print "# Your rand seems to be bogus.\n";
216 print "#\n# If you are having random number troubles,\n";
217 print "# see the hints within the test script for more\n";
218 printf "# information on why this might fail. [ %.3f ]\n", $dev;
222 srand; # These three lines are for test 7
223 my $time = time; # It's just faster to do them here.
224 my $rand = join ", ", rand, rand, rand;
228 # This test checks that the argument to srand actually
229 # sets the seed for generating random numbers.
236 print "# srand is not consistent.\n";
243 # This test just checks that the previous one didn't
244 # give us false confidence!
248 print "# rand is now unchanging!\n";
255 # This checks that srand without arguments gives
256 # different sequences each time. Note: You shouldn't
257 # be calling srand more than once unless you know
258 # what you're doing! But if this fails on your
259 # system, run perlbug and let the developers know
260 # what other sources of randomness srand should
263 while ($time == time) { } # Wait for new second, just in case.
265 if ((join ", ", rand, rand, rand) eq $rand) {
267 print "# srand without args isn't varying.\n";
273 # Now, let's see whether rand accepts its argument
276 $max = $min = rand(100);
279 $max = $n if $n > $max;
280 $min = $n if $n < $min;
285 # This test checks to see that rand(100) really falls
286 # within the range 0 - 100, and that the numbers produced
287 # have a reasonably-large range among them.
289 if ($min < 0 or $max >= 100 or ($max - $min) < 65) {
291 print "# min too low\n" if $min < 0;
292 print "# max too high\n" if $max >= 100;
293 print "# range too narrow\n" if ($max - $min) < 65;
300 # This test checks that rand without an argument
301 # is equivalent to rand(1).
303 $_ = 12345; # Just for fun.
311 print "# rand without arguments isn't rand(1)!\n";
316 # This checks that rand without an argument is not
317 # rand($_). (In case somebody got overzealous.)
321 print "# rand without arguments isn't under 1!\n";
329 # This test checks whether Perl called srand for you. This should
330 # be the case in version 5.004 and later. Note: You must still
331 # call srand if your code might ever be run on a pre-5.004 system!
335 unless ($Config{d_fork}) {
336 # Skip this test. It's not likely to be system-specific, anyway.
337 print "ok 11\n# Skipping this test on this platform.\n";
343 my $PERL = (($^O eq 'VMS') ? "MCR $^X"
344 : ($^O eq 'MSWin32') ? '.\perl'
346 $pid = open PERL, qq[$PERL -e "print rand"|];
347 die "Couldn't pipe from perl: $!" unless defined $pid;
348 if (defined $first) {
349 if ($first ne <PERL>) {
356 close PERL or die "perl returned error code $?";
358 print "not ok 11\n# srand isn't being autocalled.\n";