note
turnstep
Here's a simplified test case. For added fun, it also works
with @, $ and *:
<CODE>
#!perl
s//%/e;
}
</CODE>
<P>Eval is actually trying to use the surrounding
code to try and finish it's evaluation. This was
fixed. Recent version of perl (e.g. 5.6) will give
this warning when the code above is run:</P>
<CODE>
Bad evalled substitution pattern at file.pl line X
</CODE>
<P>Looking at perl5004delta, we read that:</P>
<CODE>
Perl (version 5) used to determine the value of
EXPR inconsistently, sometimes incorrectly using
the surrounding context for the determination. Now,
the value of EXPR (before being parsed by eval) is
always determined in a scalar context.
</CODE>
<P>Unfortunately, each version keeps removing some of these
little "features" that make for great obfuscated code. :)</P>
<P>Here's some fun code to try out if you have an "older" version of perl (older is in quotes because it hasn't been
fixed *that* recently):</P>
<CODE>
#!perl
$foo = "ABC";
print "($foo)\n";
$foo =~ s/B/*/e;
}
print "($foo)\n";
</CODE>
<P>Running this script produces this rather odd output:
<CODE>
(ABC)
(A*main::}
C)
</CODE>
<P>In other words, eval tried so hard to evaluate something, it regarded the * as a typeglob, and created
something unusual, to say the least. :)</P>
24410
24410