3 # Tests for sprintf that do not fit the format of sprintf.t.
11 # We'll run 12 extra tests (see below) if $Q is false.
12 eval { my $q = pack "q", 0 };
17 # %a and %A depend on the floating point config
18 # This totally doesn't test non-IEEE-754 float formats.
20 print "# uvsize = $Config{uvsize}\n";
21 print "# nvsize = $Config{nvsize}\n";
22 print "# nv_preserves_uv_bits = $Config{nv_preserves_uv_bits}\n";
23 print "# d_quad = $Config{d_quad}\n";
24 if ($Config{nvsize} == 8 &&
26 # IEEE-754 64-bit ("double precision"), the most common out there
27 ($Config{uvsize} == 8 && $Config{nv_preserves_uv_bits} == 53)
29 # If we have a quad we can still get the mantissa bits.
30 ($Config{uvsize} == 4 && $Config{d_quad})
34 [ '%a', '0', '0x0p+0' ],
35 [ '%a', '1', '0x1p+0' ],
36 [ '%a', '1.0', '0x1p+0' ],
37 [ '%a', '0.5', '0x1p-1' ],
38 [ '%a', '0.25', '0x1p-2' ],
39 [ '%a', '0.75', '0x1.8p-1' ],
40 [ '%a', '3.14', '0x1.91eb851eb851fp+1' ],
41 [ '%a', '-1.0', '-0x1p+0' ],
42 [ '%a', '-3.14', '-0x1.91eb851eb851fp+1' ],
43 [ '%a', '0.1', '0x1.999999999999ap-4' ],
44 [ '%a', '1/7', '0x1.2492492492492p-3' ],
45 [ '%a', 'sqrt(2)', '0x1.6a09e667f3bcdp+0' ],
46 [ '%a', 'exp(1)', '0x1.5bf0a8b145769p+1' ],
47 [ '%a', '2**-10', '0x1p-10' ],
48 [ '%a', '2**10', '0x1p+10' ],
49 [ '%a', '1e-9', '0x1.12e0be826d695p-30' ],
50 [ '%a', '1e9', '0x1.dcd65p+29' ],
52 [ '%#a', '1', '0x1.p+0' ],
53 [ '%+a', '1', '+0x1p+0' ],
54 [ '%+a', '-1', '-0x1p+0' ],
55 [ '% a', ' 1', ' 0x1p+0' ],
56 [ '% a', '-1', '-0x1p+0' ],
58 [ '%8a', '3.14', '0x1.91eb851eb851fp+1' ],
59 [ '%13a', '3.14', '0x1.91eb851eb851fp+1' ],
60 [ '%20a', '3.14', '0x1.91eb851eb851fp+1' ],
61 [ '%.4a', '3.14', '0x1.91ecp+1' ],
62 [ '%.5a', '3.14', '0x1.91eb8p+1' ],
63 [ '%.6a', '3.14', '0x1.91eb85p+1' ],
64 [ '%.20a', '3.14', '0x1.91eb851eb851f0000000p+1' ],
65 [ '%20.10a', '3.14', ' 0x1.91eb851eb8p+1' ],
66 [ '%20.15a', '3.14', '0x1.91eb851eb851f00p+1' ],
67 [ '% 20.10a', '3.14', ' 0x1.91eb851eb8p+1' ],
68 [ '%020.10a', '3.14', '0x0001.91eb851eb8p+1' ],
70 [ '%.13a', '1', '0x1.0000000000000p+0' ],
71 [ '%.13a', '-1', '-0x1.0000000000000p+0' ],
72 [ '%.13a', '0', '0x0.0000000000000p+0' ],
74 [ '%30a', '3.14', ' 0x1.91eb851eb851fp+1' ],
75 [ '%-30a', '3.14', '0x1.91eb851eb851fp+1 ' ],
76 [ '%030a', '3.14', '0x00000000001.91eb851eb851fp+1' ],
77 [ '%-030a', '3.14', '0x1.91eb851eb851fp+1 ' ],
80 '0x1.91eb851eb851f000000000000000000000000000p+1' ],
82 [ '%A', '3.14', '0X1.91EB851EB851FP+1' ],
84 } elsif (($Config{nvsize} == 16 || $Config{nvsize} == 12) &&
85 # 80-bit ("extended precision") long double, pack F is the NV
86 # cd cc cc cc cc cc cc cc fb bf 00 00 00 00 00 00
87 # cd cc cc cc cc cc cc cc fb bf 00 00
88 (pack("F", 0.1) =~ /^\xCD/ || # LE
89 pack("F", 0.1) =~ /\xCD$/)) { # BE (if this ever happens)
91 [ '%a', '0', '0x0p+0' ],
92 [ '%a', '1', '0x8p-3' ],
93 [ '%a', '1.0', '0x8p-3' ],
94 [ '%a', '0.5', '0x8p-4' ],
95 [ '%a', '0.25', '0x8p-5' ],
96 [ '%a', '0.75', '0xcp-4' ],
97 [ '%a', '3.14', '0xc.8f5c28f5c28f5c3p-2' ],
98 [ '%a', '-1.0', '-0x8p-3' ],
99 [ '%a', '-3.14', '-0xc.8f5c28f5c28f5c3p-2' ],
100 [ '%a', '0.1', '0xc.ccccccccccccccdp-7' ],
101 [ '%a', '1/7', '0x9.249249249249249p-6' ],
102 [ '%a', 'sqrt(2)', '0xb.504f333f9de6484p-3' ],
103 [ '%a', 'exp(1)', '0xa.df85458a2bb4a9bp-2' ],
104 [ '%a', '2**-10', '0x8p-13' ],
105 [ '%a', '2**10', '0x8p+7' ],
106 [ '%a', '1e-9', '0x8.9705f4136b4a597p-33' ],
107 [ '%a', '1e9', '0xe.e6b28p+26' ],
109 [ '%#a', '1', '0x8.p-3' ],
110 [ '%+a', '1', '+0x8p-3' ],
111 [ '%+a', '-1', '-0x8p-3' ],
112 [ '% a', ' 1', ' 0x8p-3' ],
113 [ '% a', '-1', '-0x8p-3' ],
115 [ '%8a', '3.14', '0xc.8f5c28f5c28f5c3p-2' ],
116 [ '%13a', '3.14', '0xc.8f5c28f5c28f5c3p-2' ],
117 [ '%20a', '3.14', '0xc.8f5c28f5c28f5c3p-2' ],
118 [ '%.4a', '3.14', '0xc.8f5cp-2' ],
119 [ '%.5a', '3.14', '0xc.8f5c3p-2' ],
120 [ '%.6a', '3.14', '0xc.8f5c29p-2' ],
121 [ '%.20a', '3.14', '0xc.8f5c28f5c28f5c300000p-2' ],
122 [ '%20.10a', '3.14', ' 0xc.8f5c28f5c3p-2' ],
123 [ '%20.15a', '3.14', '0xc.8f5c28f5c28f5c3p-2' ],
124 [ '% 20.10a', '3.14', ' 0xc.8f5c28f5c3p-2' ],
125 [ '%020.10a', '3.14', '0x000c.8f5c28f5c3p-2' ],
127 [ '%30a', '3.14', ' 0xc.8f5c28f5c28f5c3p-2' ],
128 [ '%-30a', '3.14', '0xc.8f5c28f5c28f5c3p-2 ' ],
129 [ '%030a', '3.14', '0x00000000c.8f5c28f5c28f5c3p-2' ],
130 [ '%-030a', '3.14', '0xc.8f5c28f5c28f5c3p-2 ' ],
133 '0xc.8f5c28f5c28f5c30000000000000000000000000p-2' ],
135 [ '%A', '3.14', '0XC.8F5C28F5C28F5C3P-2' ],
138 # IEEE 754 128-bit ("quadruple precision"), e.g. IA-64 (Itanium) in VMS
139 $Config{nvsize} == 16 &&
140 # 9a 99 99 99 99 99 99 99 99 99 99 99 99 99 fb 3f (LE), pack F is the NV
141 (pack("F", 0.1) =~ /^\x9A\x99{6}/ || # LE
142 pack("F", 0.1) =~ /\x99{6}\x9A$/) # BE
145 [ '%a', '0', '0x0p+0' ],
146 [ '%a', '1', '0x1p+0' ],
147 [ '%a', '1.0', '0x1p+0' ],
148 [ '%a', '0.5', '0x1p-1' ],
149 [ '%a', '0.25', '0x1p-2' ],
150 [ '%a', '0.75', '0x1.8p-1' ],
151 [ '%a', '3.14', '0x1.91eb851eb851eb851eb851eb851fp+1' ],
152 [ '%a', '-1', '-0x1p+0' ],
153 [ '%a', '-3.14', '-0x1.91eb851eb851eb851eb851eb851fp+1' ],
154 [ '%a', '0.1', '0x1.999999999999999999999999999ap-4' ],
155 [ '%a', '1/7', '0x1.2492492492492492492492492492p-3' ],
156 [ '%a', 'sqrt(2)', '0x1.6a09e667f3bcc908b2fb1366ea95p+0' ],
157 [ '%a', 'exp(1)', '0x1.5bf0a8b1457695355fb8ac404e7ap+1' ],
158 [ '%a', '2**-10', '0x1p-10' ],
159 [ '%a', '2**10', '0x1p+10' ],
160 [ '%a', '1e-09', '0x1.12e0be826d694b2e62d01511f12ap-30' ],
161 [ '%a', '1e9', '0x1.dcd65p+29' ],
163 [ '%#a', '1', '0x1.p+0' ],
164 [ '%+a', '1', '+0x1p+0' ],
165 [ '%+a', '-1', '-0x1p+0' ],
166 [ '% a', '1', ' 0x1p+0' ],
167 [ '% a', '-1', '-0x1p+0' ],
169 [ '%8a', '3.14', '0x1.91eb851eb851eb851eb851eb851fp+1' ],
170 [ '%13a', '3.14', '0x1.91eb851eb851eb851eb851eb851fp+1' ],
171 [ '%20a', '3.14', '0x1.91eb851eb851eb851eb851eb851fp+1' ],
172 [ '%.4a', '3.14', '0x1.91ecp+1' ],
173 [ '%.5a', '3.14', '0x1.91eb8p+1' ],
174 [ '%.6a', '3.14', '0x1.91eb85p+1' ],
175 [ '%.20a', '3.14', '0x1.91eb851eb851eb851eb8p+1' ],
176 [ '%20.10a', '3.14', ' 0x1.91eb851eb8p+1' ],
177 [ '%20.15a', '3.14', '0x1.91eb851eb851eb8p+1' ],
178 [ '% 20.10a', '3.14', ' 0x1.91eb851eb8p+1' ],
179 [ '%020.10a', '3.14', '0x0001.91eb851eb8p+1' ],
181 [ '%30a', '3.14', '0x1.91eb851eb851eb851eb851eb851fp+1' ],
182 [ '%-30a', '3.14', '0x1.91eb851eb851eb851eb851eb851fp+1' ],
183 [ '%030a', '3.14', '0x1.91eb851eb851eb851eb851eb851fp+1' ],
184 [ '%-030a', '3.14', '0x1.91eb851eb851eb851eb851eb851fp+1' ],
187 '0x1.91eb851eb851eb851eb851eb851f000000000000p+1' ],
189 [ '%A', '3.14', '0X1.91EB851EB851EB851EB851EB851FP+1' ],
192 # "double-double", two 64-bit doubles end to end
193 $Config{nvsize} == 16 &&
194 # bf b9 99 99 99 99 99 9a bc 59 99 99 99 99 99 9a (BE), pack F is the NV
195 (pack("F", 0.1) =~ /^\x9A\x99{5}\x59\xBC/ || # LE
196 pack("F", 0.1) =~ /\xBC\x59\x99{5}\x9A$/) # BE
200 [ '%a', '0', '0x0p+0' ],
201 [ '%a', '1', '0x1p+0' ],
202 [ '%a', '1.0', '0x1p+0' ],
203 [ '%a', '0.5', '0x1p-1' ],
204 [ '%a', '0.25', '0x1p-2' ],
205 [ '%a', '0.75', '0x1.8p-1' ],
206 [ '%a', '3.14', '0x1.91eb851eb851eb851eb851eb85p+1' ],
207 [ '%a', '-1', '-0x0p+0' ],
208 [ '%a', '-3.14', '-0x1.91eb851eb851eb851eb851eb85p+1' ],
209 [ '%a', '0.1', '0x1.999999999999999999999999998p-4' ],
210 [ '%a', '1/7', '0x1.249249249249249249249249248p-3' ],
211 [ '%a', 'sqrt(2)', '0x1.6a09e667f3bcc908b2fb1366ea8p+0' ],
212 [ '%a', 'exp(1)', '0x1.5bf0a8b1457695355fb8ac404e8p+1' ],
213 [ '%a', '2**-10', '0x1p-10' ],
214 [ '%a', '2**10', '0x1p+10' ],
215 [ '%a', '1e-09', '0x1.12e0be826d694b2e62d01511f14p-30' ],
216 [ '%a', '1e9', '0x1.dcd65p+29' ],
218 [ '%#a', '1', '0x1.p+0' ],
219 [ '%+a', '1', '+0x1p+0' ],
220 [ '%+a', '-1', '-0x1p+0' ],
221 [ '% a', '1', ' 0x1p+0' ],
222 [ '% a', '-1', '-0x1p+0' ],
224 [ '%8a', '3.14', '0x1.91eb851eb851eb851eb851eb85p+1' ],
225 [ '%13a', '3.14', '0x1.91eb851eb851eb851eb851eb85p+1' ],
226 [ '%20a', '3.14', '0x1.91eb851eb851eb851eb851eb85p+1' ],
227 [ '%.4a', '3.14', '0x1.91ecp+1' ],
228 [ '%.5a', '3.14', '0x1.91eb8p+1' ],
229 [ '%.6a', '3.14', '0x1.91eb85p+1' ],
230 [ '%.20a', '3.14', '0x1.91eb851eb851eb851eb8p+1' ],
231 [ '%20.10a', '3.14', ' 0x1.91eb851eb8p+1' ],
232 [ '%20.15a', '3.14', '0x1.91eb851eb851eb8p+1' ],
233 [ '% 20.10a', '3.14', ' 0x1.91eb851eb8p+1' ],
234 [ '%020.10a', '3.14', '0x0001.91eb851eb8p+1' ],
236 [ '%30a', '3.14', '0x1.91eb851eb851eb851eb851eb85p+1' ],
237 [ '%-30a', '3.14', '0x1.91eb851eb851eb851eb851eb85p+1' ],
238 [ '%030a', '3.14', '0x1.91eb851eb851eb851eb851eb85p+1' ],
239 [ '%-030a', '3.14', '0x1.91eb851eb851eb851eb851eb85p+1' ],
242 '0x1.91eb851eb851eb851eb851eb8500000000000000p+1' ],
244 [ '%A', '3.14', '0X1.91EB851EB851EB851EB851EB85P+1' ],
247 print "# no hexfloat tests\n";
250 plan tests => 1408 + ($Q ? 0 : 12) + @hexfloat + 12;
256 sprintf("%.40g ",0.01),
257 sprintf("%.40g", 0.01)." ",
258 q(the sprintf "%.<number>g" optimization)
261 sprintf("%.40f ",0.01),
262 sprintf("%.40f", 0.01)." ",
263 q(the sprintf "%.<number>f" optimization)
266 # cases of $i > 1 are against [perl #39126]
267 for my $i (1, 5, 10, 20, 50, 100) {
268 chop(my $utf8_format = "%-*s\x{100}");
269 my $string = "\xB4"x$i; # latin1 ACUTE or ebcdic COPYRIGHT
270 my $expect = $string." "x$i; # followed by 2*$i spaces
271 is(sprintf($utf8_format, 3*$i, $string), $expect,
272 "width calculation under utf8 upgrade, length=$i");
275 # check simultaneous width & precision with wide characters
276 for my $i (1, 3, 5, 10) {
277 my $string = "\x{0410}"x($i+10); # cyrillic capital A
278 my $expect = "\x{0410}"x$i; # cut down to exactly $i characters
279 my $format = "%$i.${i}s";
280 is(sprintf($format, $string), $expect,
281 "width & precision interplay with utf8 strings, length=$i");
284 # Used to mangle PL_sv_undef
286 'print sprintf "xxx%n\n"; print undef',
287 qr/Modification of a read-only value attempted at - line 1\./,
288 { switches => [ '-w' ] },
289 q(%n should not be able to modify read-only constants),
293 for (int(~0/2+1), ~0, "9999999999999999999") {
294 is(eval {sprintf "%${_}d", 0}, undef, "no sprintf result expected %${_}d");
295 like($@, qr/^Integer overflow in format string for sprintf /, "overflow in sprintf");
296 is(eval {printf "%${_}d\n", 0}, undef, "no printf result expected %${_}d");
297 like($@, qr/^Integer overflow in format string for printf /, "overflow in printf");
300 # check %NNN$ for range bounds
302 my ($warn, $bad) = (0,0);
303 local $SIG{__WARN__} = sub {
304 if ($_[0] =~ /missing argument/i) {
312 my $fmt = join('', map("%$_\$s%" . ((1 << 31)-$_) . '$s', 1..20));
313 my $result = sprintf $fmt, qw(a b c d);
314 is($result, "abcd", "only four valid values in $fmt");
315 is($warn, 36, "expected warnings");
316 is($bad, 0, "unexpected warnings");
319 # Tests for "missing argument" and "redundant argument" warnings
321 my ($warn_missing, $warn_redundant, $warn_bad) = (0,0,0);
322 local $SIG{__WARN__} = sub {
323 if ($_[0] =~ /missing argument/i) {
326 elsif ($_[0] =~ /redundant argument/i) {
335 # The "", "%s", and "%-p" formats have special-case handling
346 args => [ qw( x y ) ],
353 args => [ qw( x y ) ],
354 res => qr/^[0-9a-f]+$/as,
358 # Other non-specialcased patterns
361 args => [ qw( a b c ) ],
367 fmt => "%s : %s : %s",
368 args => [ qw( a b c d e ) ],
371 # Note how we'll only warn about redundant arguments once,
372 # even though both "d" and "e" are redundant...
376 fmt => "%s : %s : %s",
379 # ...But when arguments are missing we'll warn about every
380 # missing argument. This difference between the two
381 # warnings is a feature.
386 # Tests for format parameter indexes.
388 # Deciding what to do about these is a bit tricky, and so is
389 # "correctly" warning about missing arguments on them.
391 # Should we warn if you supply 4 arguments but only use
392 # argument 1,3 & 4? Or only if you supply 5 arguments and your
393 # highest used argument is 4?
395 # For some uses of this printf feature (e.g. i18n systems)
396 # it's a always a logic error to not print out every provided
397 # argument, but for some other uses skipping some might be a
398 # feature (although you could argue that then printf should be
401 # printf q[%1$s %3$s], x(), undef, z();
405 # printf q[%1$s %3$s], x(), y(), z();
407 # Since calling the (possibly expensive) y() function is
408 # completely redundant there.
410 # We deal with all these potential problems by not even
411 # trying. If the pattern contains any format parameter indexes
412 # whatsoever we'll never warn about redundant arguments.
414 fmt => '%1$s : %2$s',
415 args => [ qw( x y z ) ],
421 fmt => '%2$s : %4$s : %5$s',
422 args => [ qw( a b c d )],
428 fmt => '%s : %1$s : %s',
429 args => [ qw( x y z ) ],
437 for my $i (0..$#tests) {
438 my $test = $tests[$i];
439 my $result = sprintf $test->{fmt}, @{$test->{args}};
441 my $prefix = "For format '$test->{fmt}' and arguments/result '@{$test->{args}}'/'$result'";
442 if (ref $test->{res} eq 'Regexp') {
443 like($result, $test->{res}, "$prefix got the right result");
445 is($result, $test->{res}, "$prefix got the right result");
447 is($warn_missing, $test->{m}, "$prefix got '$test->{m}' 'missing argument' warnings");
448 is($warn_redundant, $test->{r}, "$prefix got '$test->{r}' 'redundant argument' warnings");
449 is($warn_bad, 0, "$prefix No unknown warnings");
451 ($warn_missing, $warn_redundant, $warn_bad) = (0,0,0);
456 foreach my $ord (0 .. 255) {
458 local $SIG{__WARN__} = sub {
459 if ($_[0] !~ /^Invalid conversion in sprintf/) {
464 my $r = eval {sprintf '%v' . chr $ord};
465 is ($bad, 0, "pattern '%v' . chr $ord");
469 sub mysprintf_int_flags {
470 my ($fmt, $num) = @_;
471 die "wrong format $fmt" if $fmt !~ /^%([-+ 0]+)([1-9][0-9]*)d\z/;
474 my $sign = $num < 0 ? '-' :
475 $flag =~ /\+/ ? '+' :
476 $flag =~ /\ / ? ' ' :
479 my $padlen = $width - length($sign.$abs);
481 $flag =~ /0/ && $flag !~ /-/ # do zero padding
482 ? $sign . '0' x $padlen . $abs
483 : $flag =~ /-/ # left or right
484 ? $sign . $abs . ' ' x $padlen
485 : ' ' x $padlen . $sign . $abs;
488 # Whole tests for "%4d" with 2 to 4 flags;
489 # total counts: 3 * (4**2 + 4**3 + 4**4) == 1008
491 my @flags = ("-", "+", " ", "0");
492 for my $num (0, -1, 1) {
493 for my $f1 (@flags) {
494 for my $f2 (@flags) {
495 for my $f3 ('', @flags) { # '' for doubled flags
496 my $flag = $f1.$f2.$f3;
498 my $fmt = '%'."${flag}${width}d";
499 my $result = sprintf($fmt, $num);
500 my $expect = mysprintf_int_flags($fmt, $num);
501 is($result, $expect, qq/sprintf("$fmt",$num)/);
505 for my $f4 (@flags) { # quadrupled flags
506 my $flag = $f1.$f2.$f3.$f4;
507 my $fmt = '%'."${flag}${width}d";
508 my $result = sprintf($fmt, $num);
509 my $expect = mysprintf_int_flags($fmt, $num);
510 is($result, $expect, qq/sprintf("$fmt",$num)/);
517 # test that %f doesn't panic with +Inf, -Inf, NaN [perl #45383]
518 foreach my $n (2**1e100, -2**1e100, 2**1e100/2**1e100) { # +Inf, -Inf, NaN
519 eval { my $f = sprintf("%f", $n); };
520 is $@, "", "sprintf(\"%f\", $n)";
523 # test %ll formats with and without HAS_QUAD
525 [ '%lld' => [qw( 4294967296 -100000000000000 )] ],
526 [ '%lli' => [qw( 4294967296 -100000000000000 )] ],
527 [ '%llu' => [qw( 4294967296 100000000000000 )] ],
528 [ '%Ld' => [qw( 4294967296 -100000000000000 )] ],
529 [ '%Li' => [qw( 4294967296 -100000000000000 )] ],
530 [ '%Lu' => [qw( 4294967296 100000000000000 )] ],
534 my($fmt, $nums) = @$t;
535 for my $num (@$nums) {
537 local $SIG{__WARN__} = sub { $w .= shift };
538 my $sprintf_got = sprintf($fmt, $num);
540 is($sprintf_got, $num, "quad: $fmt -> $num");
541 is($w, '', "no warnings for: $fmt -> $num");
543 is($sprintf_got, $fmt, "quad unsupported: $fmt -> $fmt");
544 like($w, qr/Invalid conversion in sprintf: "$fmt"/, "got warning about invalid conversion from fmt : $fmt");
545 like($w, qr/Redundant argument in sprintf/, "got warning about redundant argument in sprintf from fmt : $fmt");
550 # Check unicode vs byte length
551 for my $width (1,2,3,4,5,6,7) {
552 for my $precis (1,2,3,4,5,6,7) {
553 my $v = "\x{20ac}\x{20ac}";
554 my $format = "%" . $width . "." . $precis . "s";
555 my $chars = ($precis > 2 ? 2 : $precis);
556 my $space = ($width < 2 ? 0 : $width - $chars);
558 'my $v = "\x{20ac}\x{20ac}"; my $x = sprintf "'.$format.'", $v; $x =~ /^(\s*)(\S*)$/; print "$_" for map {length} $1, $2',
561 q(sprintf ").$format.q(", "\x{20ac}\x{20ac}"),
569 '""', sub { ++our $count; $_[0][0]; },
570 '0+', sub { ++our $numcount; $_[0][1]; }
572 my $o = bless ["\x{100}",42], o::;
573 () = sprintf "%1s", $o;
574 is $o::count, '1', 'sprinf %1s overload count';
576 () = sprintf "%.1s", $o;
577 is $o::count, '1', 'sprinf %.1s overload count';
579 () = sprintf "%d", $o;
580 is $o::count, 0, 'sprintf %d string overload count is 0';
581 is $o::numcount, 1, 'sprintf %d number overload count is 1';
583 my $ppc_linux = $Config{archname} =~ /^(?:ppc|power(?:pc)?)(?:64)?-linux/;
584 my $irix_ld = $Config{archname} =~ /^IP\d+-irix-ld$/;
586 for my $t (@hexfloat) {
587 my ($format, $arg, $expected) = @$t;
589 my $result = sprintf($format, $arg);
590 my $ok = $result eq $expected;
591 # For certain platforms (all of which are currently double-double,
592 # but different implementations, GNU vs vendor, two different archs
593 # (ppc and mips), and two different libm interfaces) we have some
594 # bits-in-the-last-hexdigit differences.
595 # Patch them up as TODOs instead of deadly errors.
596 if ($doubledouble && $ppc_linux && $arg =~ /^2.71828/) {
597 # gets '0x1.5bf0a8b1457695355fb8ac404ecp+1'
598 # wants '0x1.5bf0a8b1457695355fb8ac404e8p+1'
599 local $::TODO = "$Config{archname} exp(1)";
600 ok($ok, "'$format' '$arg' -> '$result' cf '$expected'");
603 if ($doubledouble && $irix_ld && $arg =~ /^1.41421/) {
604 # gets '0x1.6a09e667f3bcc908b2fb1366eacp+0'
605 # wants '0x1.6a09e667f3bcc908b2fb1366ea8p+0'
606 local $::TODO = "$Config{archname} sqrt(2)";
607 ok($ok, "'$format' '$arg' -> '$result' cf '$expected'");
611 # It seems that there can be difference in the last bits:
613 # got "0x1.5bf0a8b14576ap+1"
614 # expected "0x1.5bf0a8b145769p+1"
617 # Exact cause unknown but suspecting different fp rounding modes,
618 # (towards zero? towards +inf? towards -inf?) about which Perl
619 # is blissfully unaware.
621 # Try extracting one (or sometimes two) last mantissa
622 # hexdigits, and see if they differ in value by one.
623 my ($rh, $eh) = ($result, $expected);
625 ($_[0] =~ s/(-?0x[0-9a-fA-F]+\.)//) && return $1;
627 my $rp = extract_prefix($rh);
628 my $ep = extract_prefix($eh);
629 print "# rp = $rp, ep = $ep (rh $rh, eh $eh)\n";
630 if ($rp eq $ep) { # If prefixes match.
631 sub extract_exponent {
632 ($_[0] =~ s/([pP][+-]?\d+)//) && return $1;
634 my $re = extract_exponent($rh);
635 my $ee = extract_exponent($eh);
636 print "# re = $re, ee = $ee (rh $rh, eh $eh)\n";
637 if ($re eq $ee) { # If exponents match.
638 # Remove the common prefix of the mantissa bits.
639 my $la = length($rh);
640 my $lb = length($eh);
642 for ($i = 0; $i < $la && $i < $lb; $i++) {
643 last if substr($rh, $i, 1) ne substr($eh, $i, 1);
645 $rh = substr($rh, $i);
646 $eh = substr($eh, $i);
647 print "# (rh $rh, eh $eh)\n";
649 # If necessary, pad the shorter one on the right
650 # with one zero (for example "...1f" vs "...2",
651 # we want to compare "1f" to "20").
652 if (length $rh < length $eh) {
654 } elsif (length $eh < length $rh) {
657 print "# (rh $rh, eh $eh)\n";
658 if (length $eh == length $rh) {
659 if (abs(hex($eh) - hex($rh)) == 1) {
667 ok($ok, "'$format' '$arg' -> '$result' cf '$expected'");
670 # double-double long double %a special testing.
672 skip("uselongdouble=" . ($Config{uselongdouble} ? 'define' : 'undef')
673 . " longdblkind=$Config{longdblkind} os=$^O", 6)
674 unless ($Config{uselongdouble} &&
675 ($Config{longdblkind} == 5 ||
676 $Config{longdblkind} == 6)
677 # Gating on 'linux' (ppc) here is due to the differing
678 # double-double implementations: other (also big-endian)
679 # double-double platforms (e.g. AIX on ppc or IRIX on mips)
680 # do not behave similarly.
683 # [rt.perl.org 125633]
684 like(sprintf("%La\n", (2**1020) + (2**-1072)),
685 qr/^0x1.0{522}1p\+1020$/);
686 like(sprintf("%La\n", (2**1021) + (2**-1072)),
687 qr/^0x1.0{523}8p\+1021$/);
688 like(sprintf("%La\n", (2**1022) + (2**-1072)),
689 qr/^0x1.0{523}4p\+1022$/);
690 like(sprintf("%La\n", (2**1023) + (2**-1072)),
691 qr/^0x1.0{523}2p\+1023$/);
692 like(sprintf("%La\n", (2**1023) + (2**-1073)),
693 qr/^0x1.0{523}1p\+1023$/);
694 like(sprintf("%La\n", (2**1023) + (2**-1074)),
695 qr/^0x1.0{524}8p\+1023$/);
699 skip("negative zero not available\n", 3)
700 unless sprintf('%+f', -0.0) =~ /^-0/;
701 is(sprintf("%a", -0.0), "-0x0p+0", "negative zero");
702 is(sprintf("%+a", -0.0), "-0x0p+0", "negative zero");
703 is(sprintf("%.13a", -0.0), "-0x0.0000000000000p+0", "negative zero");
707 # [perl #127183] Non-canonical hexadecimal floats are parsed prematurely
709 skip("nv_preserves_uv_bits is $Config{nv_preserves_uv_bits}, not 53", 3)
710 unless $Config{nv_preserves_uv_bits} == 53;
713 # The 0x0.b17217f7d1cf78p0 is the original LHS value
714 # from [perl #127183], its bits are 0x162e42fefa39ef << 3,
715 # resulting in a non-canonical form of hexfp, where the most
716 # significant bit is zero, instead of one.
717 is(sprintf("%a", 0x0.b17217f7d1cf78p0 - 0x1.62e42fefa39efp-1),
719 "non-canonical form [perl #127183]");
723 no warnings 'overflow'; # Not the point here.
725 # The 0x058b90bfbe8e7bc is 0x162e42fefa39ef << 2,
726 # the 0x02c5c85fdf473de is 0x162e42fefa39ef << 1,
728 is(sprintf("%a", 0x0.58b90bfbe8e7bcp1 - 0x1.62e42fefa39efp-1),
730 "non-canonical form");
732 is(sprintf("%a", 0x0.2c5c85fdf473dep2 - 0x1.62e42fefa39efp-1),
734 "non-canonical form");