use strict;
use warnings;
[download]

at the top of your program. They'll point out many errors in your program.

Update:

$atob is -2.5
$btoc is -0.5
$atoc is -3

That's why the first elsif isn't executed: $btoc > 1 || $btoc < -1 is false.

my god I'd commented out some stuff at the top of my program as I was having trouble automating excel. And I'd done strict and warnings!!!!! I hang my head in shame and apologies for the waste of your time. Thanks to all who replied E

Adding either of the two lines to your program will make Perl tell you what goes wrong:

#!/usr/bin/perl -w

# or

use warnings;
[download]

The -w switch or use warnings; lets Perl tell you what it considers dodgy or unintended practice. For example, comparing two strings that don't look like numbers as numbers. Perl considers atob and atoc to be strings for example, which likely is not what you intended in the first place.

Adding the following line prohibits the use of barewords as strings, which even more likely prevents you from making mistakes like the one you did. In fact, it prevents so many mistakes like misspelling and reference errors, this line should be at the top of all your Perl programs:

use strict;
[download]

With that line, Perl tells you that:

Bareword "atob" not allowed while "strict subs" in use at ...
[download]

which is an immediate hint. Either declare atob as a subroutine, or use $atob, as you propably intended.

print join ' ', $atob, $btoc, $atoc, "\n";
[download]

For clarity, I'd also use the abs function rather than nested logic:

if (abs $atob > 1  &&  
    abs $btoc > 1  &&  
    abs $atoc > 1     ) {

    print OUTFILE "DIFF > 1 all three\n";
    $matrixflags[$row - 4][$i] = 1;
    $matrixflags[$row - 3][$i] = 1; 
    $matrixflags[$row - 2][$i] = 1;
}

elsif (abs $atob > 1  &&
       abs $btoc < 1  &&
       abs $atoc > 1    ) {

    print OUTFILE "DIFFA > 1 $atob  $atoc $btoc\n";
    $matrixflags[$row - 4][$i] = 1;
    $matrixflags[$row - 3][$i] = 0;
    $matrixflags[$row - 2][$i] = 0;
}
elsif ( ... ) { ... }
[download]

print join ' ', $atob, $btoc, $atoc, "\n";
[download]

Why use join? It's far simpler to interpolate inside double-quotes.

print qq{$atob $btoc $atoc\n};
[download]

Cheers,

JohnGG

Update: Corrected sigil typo

I use join like this all over the place. In many cases, I find that it looks less cluttered in the code. There's no great advantage to it in this case, but there's no reason to avoid it either.

If your /[abc]to[abc]/ vars are intended to describe absolute distances, I would recommend going with the abs() value as previously recommended, then using >=1 or <=1 for one of the two cases. You probably haven't seen it yet if your data is really real, but it could burn you in the future.

It looks like you're overcomplicating things. By looking over your code, it looks like you really want something like this (untested!):

$a= 20;
$b= 22.5;
$c= 23
$atob = abs($a - $b) > 1 ? 1 : 0;
$btoc = abs($b - $c) > 1 ? 1 : 0;
$atoc = abs($a - $c) > 1 ? 1 : 0;

if ( $atob && $atoc && $btoc ) {
print OUTFILE "DIFF > 1 all three\n";
    $matrixflags[$row - 4][$i] = 1;
    $matrixflags[$row - 3][$i] = 1;
    $matrixflags[$row - 2][$i] = 1;
}
...
[download]

((btoc < 1) && (btoc > -1))
[download]

...roboticus

Update: After another second of thought, perhaps this would be even more to your liking--Replace the if .. then stuff with:

if ($atob+$atoc+$btoc) {
    printf OUTFILE "DIFF=%d:   %f,  %f,  %f\n",
        $atob+$atoc+$btoc, $a-$b, $a-$c, $b-$c;
}
$matrixflags[$row - 4][$i] = $atob;
$matrixflags[$row - 3][$i] = $atoc;
$matrixflags[$row - 2][$i] = $atob;
}
[download]

First of all, that isn't your code -- here's what perltidy and the Perl compiler would like it to be:

#!/usr/bin/perl -w

use strict;

{
    my $a    = 20;
    my $b    = 22.5;
    my $c    = 23;
    my $atob = $a - $b;
    my $btoc = $b - $c;
    my $atoc = $a - $c;
    my @matrixflags;

    if ((($atob > 1) || ($atob < -1)) && (($atoc > 1) ||     
         ($atoc < -1))
          && (($btoc > 1) || ($btoc < -1)))
    {
        print OUTFILE "DIFF > 1 all three\n";
        $matrixflags[$row - 4][$i] = 1;
        $matrixflags[$row - 3][$i] = 1;
        $matrixflags[$row - 2][$i] = 1;
    } elsif ((($atob > 1) || ($atob < -1)) && (($btoc < 1)   
           && ($btoc > -1))
          && (($atoc > 1) || ($atoc < -1)))
    {
        print OUTFILE "DIFFA > 1 $atob  $atoc $btoc\n";
        $matrixflags[$row - 4][$i] = 1;
        $matrixflags[$row - 3][$i] = 0;
        $matrixflags[$row - 2][$i] = 0;
    } elsif ((($atob > 1) || ($atob < -1)) && (($btoc > 1) 
           || ($btoc < -1))
          && (($atoc < 1) && ($atoc > -1)))
    {
        print OUTFILE "DIFFB > 1 $atob  $atoc $btoc\n";
        $matrixflags[$row - 4][$i] = 0;
        $matrixflags[$row - 3][$i] = 1;
        $matrixflags[$row - 2][$i] = 0;
    } elsif ((($atob < 1) && ($atob > -1)) && (($btoc > 1) 
           || ($btoc < -1))
          && (($atoc > 1) || ($atoc < -1)))
    {
        print OUTFILE "DIFFC > 1 $atob  $atoc $btoc\n";
        $matrixflags[$row - 4][$i] = 0;
        $matrixflags[$row - 3][$i] = 0;
        $matrixflags[$row - 2][$i] = 1;
    }

    #So in this example in my head the 1st elsif should be executed.
    #Difference $atob = 1.5 so (($atob > 1) || ($atob < -1)) is true.
    #Difference $btoc = 0.5 so (($btoc < 1) && ($btoc > -1)) is true.
    #Difference $atoc = 3 so (($atoc > 1) || ($atoc < -1)) is true.
}
[download]

Secondly, even that throws errors -- neither $row or $i are defined (either using my or programmatically). Why not show us a fragment of code that at least compiles and runs, then we'll be able to help you.