Re^2: Sort never returns data in right order

[afoken]

Just a few comments on that piece of code:

  # remove extra spaces
  my $fields = () = $summary =~ /[\s+,:]/g;

Misleading comment, misleading variable name. No space is removed from anywhere. It's just attempting to count the number of field separators, and it will probably fail. \s+ looks like you want to match any number of spaces, but that won't happen:

>perl -MYAPE::Regex::Explain -E 'say YAPE::Regex::Explain->new(q<[\s+,
+:]>)->explain'
The regular expression:

(?-imsx:[\s+,:])

matches as follows:

NODE                     EXPLANATION
----------------------------------------------------------------------
(?-imsx:                 group, but do not capture (case-sensitive)
                         (with ^ and $ matching normally) (with . not
                         matching \n) (matching whitespace and #
                         normally):
----------------------------------------------------------------------
  [\s+,:]                  any character of: whitespace (\n, \r, \t,
                           \f, and " "), '+', ',', ':'
----------------------------------------------------------------------
)                        end of grouping
----------------------------------------------------------------------

And by the way: The split pattern does not fit the input.

  elsif ( $fields == 10 ) {
    ( $Star, $Intf, $IHQ, $IQD, $OHQ, $OQD, $RXBS, $RXPS, $TXBS, $TXPS
+, $TRTL )
      = split( ' ', $summary );

$fields was calculated for a different set of field separators. Additionally, split on ' ' is special cased to emulate awk, see split.

This is overly complex. Just split the current line into an array, check if the array has the expexted number of fields (@array==10), and go on from there.

  my $Star = 0;
  my $Intf = 0;
  my $IHQ  = 0;
  my $IQD  = 0;
  my $OHQ  = 0;
  my $OQD  = 0;
  my $RXBS = 0;
  my $RXPS = 0;
  my $TXBS = 0;
  my $TXPS = 0;
  my $TRTL = 0;

  my $Track = "";

  # ...
  } elsif ( $fields == 10 ) {
    ( $Star, $Intf, $IHQ, $IQD, $OHQ, $OQD, $RXBS, $RXPS, $TXBS, $TXPS
+, $TRTL )
      = split( ' ', $summary ); 

  # ...
      $Track = join "<<-", $rec, $Intf;
      $interface_bytes{$Track} += $RXBS;
      $Track = join "->>", $rec, $Intf;
      $interface_bytes{$Track} += $TXBS;
  # ...
  }

Scope of the variables should be limited to the block following elsif, i.e. my ($Star, $Intf, ...) = split .... Assigning unused fields to write-only variables is not needed, use undef instead: my ($x,undef,$y,undef,$z)=split .... Changing the code to split into an array instead of guessing field separators would require changes here, you would use just a constant index into an array. Readonly and constant could help avoiding magic numbers for the indexes, but on the other hand, you need those field numbers only here.

      $Track = join "<<-", $rec, $Intf;
      $interface_bytes{$Track} += $RXBS;

join is overkill here. Just use string interpolation. That also gets rid of the $Track variable:

      $interface_bytes{"$rec<<-$Intf"}+=$RXBS;

    if    ( $Star ne "*" )  { next; }
    elsif ( $RXBS =~ /\D/ ) { next; }
    elsif ( $TXBS =~ /\D/ ) { next; }

You check for errors, but you don't report them. Why?

Yes, I see that the heading line will trigger those errors. But why don't you get rid of the header line before working with the input?

  } else {
    print STDERR
"Danger Will Robinson - my sensors detect an invisible hole that may c
+onsume you\n";
  }

That perfectly explains the problem. For every f*ing line. Imagine reading in 10k lines from the wrong file. Seeing the same lame joke 10_000 times is not funny at all. If you find an unrecoverable error, just die, with a reasonable error message!

---

I'm sure I would find more things that I don't like if I would take some time to actually review the code. But to summarize:

Don't post bad code!

Alexander

[stall]

I believe the previous person's post missed the point of my posts and does a disservice to the Perl community.