There's probably more than one way to accomplish the goal in the title -- and I'd love to hear stories about it, if only because it'd make me feel better.

I had the following, very simple line of code: 

$user = ~ s/^\s+|\s+$//gs;

[download]

The intention, of course, was to trim white space from the beginning and end of a user name previously parsed from a record. Instead, though, $user ended up containing a large and random-seeming number.

Spot the bug?

Here's a hint from the Perl man pages:
Unary "~" performs bitwise negation, i.e., 1's complement. For example, 0666 & ~027 is 0640. (See also "Integer Arithmetic" and "Bitwise String Operators".) Note that the width of the result is platform-dependent: ~0 is 32 bits wide on a 32-bit platform, but 64 bits wide on a 64-bit platform, so if you are expecting a certain bit width, remember to use the & operator to mask off the excess bits. -- Symbolic Unary Operators

Yes, a single space between my = and my ~ resulted in my intended "match" operation instead assigning a numeric value to $user. $head .= $desk

Stepping through the code with the debugger found the line where the oddness was occurring, but it seemed like inexplicable madness (for something like 10 minutes) until I noticed that one extra space. Man, do I feel stupid!

The lesson? I probably should have used Text::Trim. :)

<–radiant.matrix–>
Ramblings and references
The Code that can be seen is not the true Code
I haven't found a problem yet that can't be solved by a well-placed trebuchet

Replies are listed 'Best First'.
Re: How to introduce a frustrating bug with a single whitespace
by ikegami (Patriarch) on Dec 24, 2007 at 20:17 UTC
    In a similar vein, 
    print (4+5)*20;  # Prints 9, returns 20

    [download]
[reply]
[d/l]

      Yes, but at least that gets caught with warnings (or -w): 

      $ perl -we 'print (4+5)*20;'
print (...) interpreted as function at -e line 1.
Useless use of multiplication (*) in void context at -e line 1.

      [download]

      The whitespace bug described in the original node is the first bug I've had in a while that:

      • wasn't caught by syntax checking (it's valid Perl)
      • wasn't griped about with strict and warnings on
      • wasn't a design error
      <–radiant.matrix–>
      Ramblings and references
      The Code that can be seen is not the true Code
      I haven't found a problem yet that can't be solved by a well-placed trebuchet
[reply]
[d/l]
        Well, it's kind of you to say that, but given the fact that Perl 6 changed =~ to ~~, it arguably was a design error. (Though this particular failure mode is not why it was changed, but rather the failure of reversing the operator to say ~= instead. It doesn't matter if you reverse ~~.) Anyway, if you split ~~ in Perl 6 with whitespace, it'll parse, but will likely give you a "useless use of ~ in void context" warning.

        By the way, Perl 6 also fixes the "print (4+5)*20" faq...

[reply]
[d/l]
[select]

        The whitespace bug described in the original node is the first bug I've had in a while that:
        • wasn't caught by syntax checking (it's valid Perl)
        • wasn't griped about with strict and warnings on
        • wasn't a design error

        I've got one of those, though not a whitespace bug. Just recently I was messing with a data-structure that was supposed to be an array of arrays of arrays: 

        $rect_list = 
  [  [ [0, 0], [3, 9 ] ],
     [ [5, 4], [3, 10] ], 
     [ [7, 8], [9, 11] ]. 
     [ [2, 4], [5, 15] ], 
     [ [1, 9], [9, 13] ], 
   ];

        [download]

        First question: Can you spot the bug? Second question: Do you know what it does?

[reply]
[d/l]
Re: How to introduce a frustrating bug with a single whitespace
by ambrus (Abbot) on Dec 25, 2007 at 09:24 UTC

    Did you know that <$x> is a readline, but <$ x> is a glob?

[reply]
[d/l]
[select]
Re: How to introduce a frustrating bug with a single whitespace
by doom (Deacon) on Dec 31, 2007 at 02:44 UTC

    The lesson? I probably should have used Text::Trim. :)

    It could be, but I would say the lesson is to not try to do it all in one step. I remember reading a Randal Schwartz rant some time back about people continually trying to do whitespace trimming in one step, when it's much simpler and straight-forward to do it in two: 

    $user =~ s/^\s+//s;
$user =~ s/\s+$//s;

    [download]
    I think you'll find that that runs much faster than one regexp involving alternation....

[reply]
[d/l]

      I think you'll find that that runs much faster than one regexp involving alternation....

      Well, I don't really care much about faster runtimes -- most of what I do waits on disk access and network I/O long before it runs into RAM or CPU limits, so which regex is fastest almost never comes into play.

      I try to write primarily for readability. I realize, of course, that "readability" is subjective. But, I find s{^\s+|\s+$}{}g to be more readable than the two-line alternative. When I see that one line, I immediately think "ok, trim whitespace from the ends". When I see the two, I have to stop and think about it. Just me.

      One of the things I like about Python1 is the readability of performing this action: 

      text = " a string of some sort, with space at either end   "
text = text.strip()

      [download]

      I think that's what draws me to Text::Trim (if I have to do the step repeatedly, especially): 

      my $text = " a string of some sort, with space at either end   ";
$text = trim( $text );

      [download]

      Of course, creating a dependency for something like that seems a little excessive, so I've recently taken to 

      my $TRIM_WHITESPACE_RE = qr/^\s+|\s+$/
## ... later on .. ##
$text =~ s/$TRIME_WHITESPACE_RE//g;

      [download]

      1: I know, I know. :) Seriously, though, Python has some nifty things in it, even if I still like Perl better.

      <–radiant.matrix–>
      Ramblings and references
      The Code that can be seen is not the true Code
      I haven't found a problem yet that can't be solved by a well-placed trebuchet
[reply]
[d/l]
[select]

Back to Meditations