in reply to Re^2: fall through switch/case in perl
in thread fall through switch/case in perl

The eval isn't required (unless I missed a subtlety?) 

#! perl -sw
use strict;

my $var = 7;

goto 'CASE'.($var+0); die "$var out of range";

CASE10: print "a";
CASE9: print "b";
CASE8: print "c";
CASE7: print "d";
CASE6: print "e";
CASE5: print "f";
CASE4: print "g";
CASE3: print "h";
CASE2: print "i";
CASE1: print "j";

print "\n";

__END__
P:\test>junk
defghij

[download]
Examine what is said, not who speaks.
"Efficiency is intelligent laziness." -David Dunham
"Think for yourself!" - Abigail
"Memory, processor, disk in that order on the hardware side. Algorithm, algorithm, algorithm on the code side." - tachyon

Replies are listed 'Best First'.
Re^4: fall through switch/case in perl
by etcshadow (Priest) on Sep 07, 2004 at 02:52 UTC
    The eval is necessary if you want to fully emulate the C semantic of not dieing if there is no matching case. Really, the full-on computed goto method (I should have just written this into my top-level response) is this (I think): 
    {
eval { goto "CASE$var" } or goto DEFAULT;

    CASE10: print "a";
    CASE9:  print "b";
    CASE8:  print "c";
    CASE7:  print "d";
    CASE6:  print "e";
    CASE5:  print "f";
    CASE4:  print "g";
    CASE3:  print "h";
    CASE2:  print "i";
    CASE1:  print "j";
    DEFAULT: 
}

    [download]
    Which, admittedly, is not very different from Aristotle's. The only real difference being the outter braces. The purpose of them is to avoid polluting the label name-space. (Also, I prefer the "goto X or goto DEFAULT" over the "goto X; goto DEFAULT" purely for aesthetic reasons. :-D)

    I sort of wonder why this isn't given as one of the ways of achieving C-like switch statement behavior in the perl docs? Oh, well... it's probably horribly inefficient or something (apart from just being too C-ish or something). 

    ------------
:Wq
Not an editor command: Wq

    [download]
[reply]
[d/l]
[select]

      Hmm. I think what your saying is that the eval is necessary to avoid a warning if $var doesn't contain a value for which there is a label defined?

      If so, then it still doesn't handle the case where $var is undefined.

      That can be handled by using a do BLOCK construct to disable warnings and adding a CASE0: 

      for my $var ( undef, 1 .. 10, 'fred' ) {
    goto 'CASE' . do{ local $^W; $var + 0 };

    CASE10: print "a";
    CASE9:  print "b";
    CASE8:  print "c";
    CASE7:  print "d";
    CASE6:  print "e";
    CASE5:  print "f";
    CASE4:  print "g";
    CASE3:  print "h";
    CASE2:  print "i";
    CASE1:  print "j";
    print "\n";
    CASE0:
}

      [download]
      I sort of wonder why this isn't given as one of the ways of achieving C-like switch statement behavior in the perl docs? Oh, well... it's probably horribly inefficient or something (apart from just being to C-ish or something).

      C-ish it may be (though that's not a good reason in my book if it is clearer), but inefficient it is not: 

      P:\test>388915
?> 1..10
Testing for vals [1..10]
               Rate  if_cascade and_cascade     goto_do   goto_eval   
+     goto
if_cascade  11395/s          --         -2%        -11%        -26%   
+     -26%
and_cascade 11627/s          2%          --         -9%        -24%   
+     -25%
goto_do     12845/s         13%         10%          --        -17%   
+     -17%
goto_eval   15389/s         35%         32%         20%          --   
+      -0%
goto        15423/s         35%         33%         20%          0%   
+       --
?>

      [download]

      Benchmark/results

      Read more... (10 kB)
      Examine what is said, not who speaks.
      "Efficiency is intelligent laziness." -David Dunham
      "Think for yourself!" - Abigail
      "Memory, processor, disk in that order on the hardware side. Algorithm, algorithm, algorithm on the code side." - tachyon
[reply]
[d/l]
[select]
        Hmm. I think what your saying is that the eval is necessary to avoid a warning if $var doesn't contain a value for which there is a label defined?
        No, I'm saying that goto $label dies if the label has not been defined. That is not how it works in C. In C, if you have not provided a case that matches your switch value, it jumps to the default case. That's why there's an eval, to prevent the die (not warning). 
        [sstone@granite sstone]$ perl -le 'print 1; goto "FOO"; print 2'
1
Can't find label FOO at -e line 1.
[sstone@granite sstone]$

        [download]
        As far as preventing a warning on undef, well... I'd imagine that that should fit into whatever scheme you are using around undefs: warn if you have undef warnings on, else not. And there's nothing about stringifying $var that changes that semantic, so all seems well and good with my method. 
        ------------
:Wq
Not an editor command: Wq

        [download]
[reply]
[d/l]
[select]

In Section Seekers of Perl Wisdom