bradcathey has asked for the wisdom of the Perl Monks concerning the following question:

I do a lot of conditional testing, such as: 
        if     ($x == 1) { $abc = 1 }
        elsif ($x == 2) { $def = 1 }
        elsif ($x == 3) { $xyz = 1 }
        else                  { $nothing = 1 }

[download]
but in reading Prog. Perl (p. 125) I see this could send me to the confessional. So, I tried: 
    SWITCH: {
        if ($x == 1)    { $abc = 1; last SWITCH; };
        if ($x == 2)    { $def = 1; last SWITCH; };
        if ($x == 3)    { $xyz = 1; last SWITCH; };
        $nothing = 1;
    }

[download]
or 
    SWITCH: for ($x) {
        if (1)    { $abc = 1; last SWITCH; };
        if (2)    { $def = 1; last SWITCH; };
        if (3)    { $xyz = 1; last SWITCH; };
        $nothing = 1;
    }

[download]
Sure it looks nicer, but in speed tests, depending on how $x is set, this later structure always slower, and in some cases significantly.

My question, what am I missing here? The if/else structure actually looks cleaner to me and if it's faster...

Thanks, monks of greater XP.

bradcathey

Replies are listed 'Best First'.
Re: Use of SWITCH vs IF ELSE in speed and readability
by CombatSquirrel (Hermit) on Aug 30, 2003 at 16:07 UTC
    bradcathey, your fist piece of code is perfectly alright, the second one is an emulated switch statement and the third one is wrong. It would have to be if ($_ == 1). Just execute the following code and you'll see it: 
    $x = 3;
SWITCH: for ($x) {
    if (1)    { print '1'; last SWITCH; };
    if (2)    { print '2'; last SWITCH; };
    if (3)    { print '3'; last SWITCH; };
    print 'caught';
}

    [download]
    It's going to print 1, because 1 is true, so that the if (1) statement is always executed.
    Hope this helped.
    CombatSquirrel.
    Entropy is the tendency of everything going to hell.
[reply]
[d/l]
[select]
      Thanks CombatSquirrel, your answers are always helpful, and innovative.

      So, there is nothing wrong with using the if/else structure? Just don't know why the Perl book says "horrors!"
[reply]
        You know, TIMTOWDI. And I personally prefer the if/elsif/else structures - maybe I'm a bit too much of a C(++) programmer ;-)
        Cheers,
        CombatSquirrel.
        Entropy is the tendency of everything going to hell.
[reply]
Re: Use of SWITCH vs IF ELSE in speed and readability
by Zaxo (Archbishop) on Aug 30, 2003 at 16:29 UTC

    Another approach is to use a dispatch table - a hash of code references, 

    my $dtbl = {
    1 => sub { $abc = 1 },
    2 => sub { $def = 1 },
    3 => sub { $xyz = 1 },
};

exists $dtbl->{$x} ? $dtbl->{$x}() : $nothing = 1;

    [download]

    After Compline,
    Zaxo

[reply]
[d/l]
      Just a style difference, but I really dislike using the ternary ? : for statements (code which is of irrelevant or void context, performed for their side-effects). It's all just code and doesn't matter inside the machine, but to me, the ternary is intended to express a selection of either one expression or the other.

      --
      [ e d @ h a l l e y . c c ]

[reply]
[d/l]
Re: Use of SWITCH vs IF ELSE in speed and readability
by perrin (Chancellor) on Aug 30, 2003 at 18:06 UTC
    Unless it's in a critical section of code, use whichever one seems most readable in your context. It's a tough call in this example, but sometimes I find the regex approach (what you were trying to do on the 3rd one) the most readable. When speed really counts, just use the elsifs. They're not so bad.
[reply]
Re: Use of SWITCH vs IF ELSE in speed and readability
by BrowserUk (Patriarch) on Aug 31, 2003 at 08:03 UTC

    For simple conditions and small blocks of code, there's not too much wrong with an if/elsif cascade, except perhaps that if all your conditions are integers, especially a contiguous range of integers, it means that you are sometimes testing more conditions than necessary.

    For small, contiguous ranges of conditions, I prefer a array-based dispatch table. This can also be coded inline, although having the condition at the end may be seen as a disadvantage: 

    ( 
    sub{ $abc == 1 },
    sub{ $def == 1 },
    sub{ $xyz == 1 },
)[ $x - 1 ]->();

    [download]

    Somewhat obfuscate, but quite clear and efficient for small contiguous ranges. The absence of a simple 'default' mechanism is a shame.

    For discontiguous ranges, a similar arrangment using an anonymous hash can work 

    ${{
    1 => sub{ $abc == 1 },
    3 => sub{ $def == 1 },
    5 => sub{ $xyz == 1 },
}}{ $x }->();

    [download]

    The biggest problem, besides the obscurity of the construction, is the absence of a simple 'default' mechanism.

    Of course they don't have to be inline, and can used a named array or hash, which also gives the possibility for providing for a default.

    Examine what is said, not who speaks.
    "Efficiency is intelligent laziness." -David Dunham
    "When I'm working on a problem, I never think about beauty. I think only how to solve the problem. But when I have finished, if the solution is not beautiful, I know it is wrong." -Richard Buckminster Fuller
    If I understand your problem, I can solve it! Of course, the same can be said for you.
[reply]
[d/l]
[select]
Re: Use of SWITCH vs IF ELSE in speed and readability
by TomDLux (Vicar) on Aug 30, 2003 at 18:28 UTC

    If you're really comparing based on small integers, and really modifying the value of variables in a similar way, I would consider using an array of scalar references .... or re-writing to use an array or hash to replace the scalars:

    my @table = ( \$abc, \$def, \$wyz );
$table[$x] = 1;
# or
my @categories;
$categories[$x]++;
# or
my %categories;
$categories{$x} = code2value( $x );

    [download]

    Not a religious thing, but it does have the advantage of automatically handling new categories, if you can arrange for the returned values to remain small integers.

    --
    TTTATCGGTCGTTATATAGATGTTTGCA

[reply]
[d/l]

Back to Seekers of Perl Wisdom