On Sat, Feb 08, 2003 at 09:05:14PM +0200, Enache Adrian wrote:
> It fixes #20154, #20357, #19061 and passes all tests.
Is there anything wrong with my patch ? I use it myself since
a week and saw no problems so far.
( I agree with Stephen McCamant analyse/conclusion in the
redo case, but my solution - pushing the first OP_LEAVESUB,
still looks fine - I see no other way to force the execution
of the labeled COP ).
Anyway, here is the sh script turned into a regression script:
#! /usr/bin/perl
require "test.pl";
is ( runperl( prog => '{ goto A; A: print "a" } continue { $s++ }' ),
'a', 'goto inside /{ } continue { }/ loop');
is ( runperl( prog => '{ $s++ } continue { goto A; A: print "a" }' ),
'a', 'goto inside /do { } continue { }/ loop');
is ( runperl( prog =>
'foreach(1) { goto A; A: print "a" } continue { $s++ }' ),
'a', 'goto inside /foreach () { } continue { }/ loop');
is ( runperl( prog => '
sub a {
A: {
if ($p) {
redo A; B: print "a"; redo A;
}
}
goto B unless $r++
}
a();print "b"'), 'ab', 'loop label wiped away by goto');
# this works in 5.8.0, it should continue to work
is ( runperl( prog => 'for ($p=1;$p && goto A;$p=0) { A: print "a" }' ),
'a', 'wierd case of goto and for(;;) loop');
__END__
Regards
Adi
