From jandrea@grassroots.ns.ca

perl -v shows
"This is perl, version 5.004_04 built for i386-linux"
on Debian GNU/Linux 2.1
kernelversion reports 2.0
on a fairly new Pentium(tm).

Code fails with many errors in BigFloat module,
"perl -w pi-big.pl"
shows many lines of

Use of uninitialized value at /usr/lib/perl5/Math/BigFloat.pm line 78.
Use of uninitialized value at /usr/lib/perl5/Math/BigFloat.pm line 96.
Use of uninitialized value at /usr/lib/perl5/Math/BigInt.pm line 189.
Use of uninitialized value at /usr/lib/perl5/Math/BigInt.pm line 259.

Sorry if the problem is fixed in 5_005, I tried searching the PerlbugDB via
the web, but it only reports "found 0, request failed".

#!/usr/local/bin/perl

use Math​::BigFloat;

$scale = 10;

$zero = new Math​::BigFloat "0.0";
$one = new Math​::BigFloat "1.0";
$two = new Math​::BigFloat "2.0";
$four = new Math​::BigFloat "4.0";
$root2 = $two->fsqrt( $scale );

print "$zero\n";
print "$one\n";
print "$two\n";
print "$four\n";
print "$root2\n";

$a = $one;
$x = $one;
$b = $one->fdiv( $root2, $scale );
$c = $one->fdiv( $four, $scale );
$y = $one;

print "$b\n";
print "$c\n";

$maxloop = 3;
$i = 0;

$prev = $zero;
$t = $one;
$t1 = $one;
$t2 = $one;

while( $i < $maxloop ) {
  $prev = $y;
  $y = $a;
  $t = $a * $b;
  $a = $t->fdiv( $two, $scale );
  $t = $b * $y;
  $b = $t->fsqrt( $scale );
  $t = $a - $y;
  $c = $c - $x * $t * $t;
  $x = $x * $two;
  $t = $a + $b;
  $t1 = $t * $t;
  $t2 = $c * $four;
  $y = $t1->fdiv( $t2, $scale );

# $t = $y->fround( $scale );
# $y = $y->fround( $scale );

  $i++;
  print "$y\n";
}

#eof

From @floatingatoll

This bug has changed slightly; running the test script with -w shows
a series of undefined warnings from line 96 of Math/BigFloat.pm.
Before, there were several warnings - so it's mostly fixed.

R.

[some snipping]

Sourceaddr​: John Andrea <jandrea@​grassroots.ns.ca>

Just goofing around and tried using BigFloat to calculate Pi to
lots of decimal places (code at very bottom of message).

perl -v shows
"This is perl, version 5.004_04 built for i386-linux"
on Debian GNU/Linux 2.1 kernelversion reports 2.0
on a fairly new Pentium(tm).

Code fails with many errors in BigFloat module,
"perl -w pi-big.pl" shows many lines of

Use of uninitialized value at
/usr/lib/perl5/Math/BigFloat.pm line 96.

[snip]

#!/usr/local/bin/perl

use Math​::BigFloat;

$scale = 10;

$zero = new Math​::BigFloat "0.0";
$one = new Math​::BigFloat "1.0";
$two = new Math​::BigFloat "2.0";
$four = new Math​::BigFloat "4.0";
$root2 = $two->fsqrt( $scale );

print "$zero\n";
print "$one\n";
print "$two\n";
print "$four\n";
print "$root2\n";

$a = $one;
$x = $one;
$b = $one->fdiv( $root2, $scale );
$c = $one->fdiv( $four, $scale );
$y = $one;

print "$b\n";
print "$c\n";

$maxloop = 3;
$i = 0;

$prev = $zero;
$t = $one;
$t1 = $one;
$t2 = $one;

while( $i < $maxloop ) {
$prev = $y;
$y = $a;
$t = $a * $b;
$a = $t->fdiv( $two, $scale );
$t = $b * $y;
$b = $t->fsqrt( $scale );
$t = $a - $y;
$c = $c - $x * $t * $t;
$x = $x * $two;
$t = $a + $b;
$t1 = $t * $t;
$t2 = $c * $four;
$y = $t1->fdiv( $t2, $scale );

# $t = $y->fround( $scale );
# $y = $y->fround( $scale );

$i++;
print "$y\n";
}

#eof

From @JohnPeacock

Check this again with the next build that Jarkko sends out. He recently
applied the patch I submitted which should fix all of those problems. I
see no more warnings running with my patched Math​::BigFloat and 5.6.0.

John Peacock

Richard Soderberg wrote​:

This bug has changed slightly; running the test script with -w shows
a series of undefined warnings from line 96 of Math/BigFloat.pm.
Before, there were several warnings - so it's mostly fixed.

R.

[some snipping]

Sourceaddr​: John Andrea <jandrea@​grassroots.ns.ca>

Just goofing around and tried using BigFloat to calculate Pi to
lots of decimal places (code at very bottom of message).

perl -v shows
"This is perl, version 5.004_04 built for i386-linux"
on Debian GNU/Linux 2.1 kernelversion reports 2.0
on a fairly new Pentium(tm).

Code fails with many errors in BigFloat module,
"perl -w pi-big.pl" shows many lines of

Use of uninitialized value at
/usr/lib/perl5/Math/BigFloat.pm line 96.

[snip]

#!/usr/local/bin/perl

use Math​::BigFloat;

$scale = 10;

$zero = new Math​::BigFloat "0.0";
$one = new Math​::BigFloat "1.0";
$two = new Math​::BigFloat "2.0";
$four = new Math​::BigFloat "4.0";
$root2 = $two->fsqrt( $scale );

print "$zero\n";
print "$one\n";
print "$two\n";
print "$four\n";
print "$root2\n";

$a = $one;
$x = $one;
$b = $one->fdiv( $root2, $scale );
$c = $one->fdiv( $four, $scale );
$y = $one;

print "$b\n";
print "$c\n";

$maxloop = 3;
$i = 0;

$prev = $zero;
$t = $one;
$t1 = $one;
$t2 = $one;

while( $i < $maxloop ) {
$prev = $y;
$y = $a;
$t = $a * $b;
$a = $t->fdiv( $two, $scale );
$t = $b * $y;
$b = $t->fsqrt( $scale );
$t = $a - $y;
$c = $c - $x * $t * $t;
$x = $x * $two;
$t = $a + $b;
$t1 = $t * $t;
$t2 = $c * $four;
$y = $t1->fdiv( $t2, $scale );

# $t = $y->fround( $scale );
# $y = $y->fround( $scale );

$i++;
print "$y\n";
}

#eof