My appologies - where I wrote utf.pm, I should have written utf8.pm.
Running under the Perl debugger with trace on, ends as follows:
main::(test4.pl:17): $text =~ m![t]*!;
main::CDE(0x95c3d68)(test4.pl:17):
17: $text =~ m![t]*!;
Encode::find_encoding(/):
121: my ($name, $skip_external) = @_;
Encode::find_encoding(/):
122: return __PACKAGEgetEncoding($name,$skip_external);
Encode::getEncoding(/):
96: my ($class, $name, $skip_external) = @_;
Encode::getEncoding(/):
98: ref($name) && $name->can('renew') and return $name;
Encode::getEncoding(/):
99: exists $Encoding{$name} and return $Encoding{$name};
Encode::Unicode::renew(/ 6):
46: my $self = shift;
Encode::Unicode::renew(/ 7):
47: $BM_Unknown{$self->name} or return $self;
Encode::Encoding::name(/ 18):
18: sub name { return shift->{'Name'} }
Encode::Unicode::renew(/ 8):
48: my $clone = bless { %$self } =ref($self);
Encode::Unicode::renew(/ 9):
49: $clone->{renewed}++; # so the caller knows it is renewed.
Encode::Unicode::renew(/ 0):
50: return $clone;
Encode::Encoding::needs_lines(/ ing.pm:34):
34: sub needs_lines { 0 };
Encode::Encoding::DESTRY(/ pm:64):
64: sub DESTRY {}
UTF-16:Unrecognised BM 7061, <INline 1.
at / line 0
require utf8.pm called at test4.pl line 17
main::BEGIN() called at / line 0
eval {} called at / line 0
Compilation failed in require at test4.pl line 17, <INline 1.
BEGIN failed aborted, <INline 1.
at test4.pl line 0
Debugged program terminated. Use q to quit or R to restart,

Sun, 20 Aug 2006 15:23:11 +1200, <ian.goodacre (AT) xtra (DOT) co.nzwrote

My appologies - where I wrote utf.pm, I should have written utf8.pm.

Running under the Perl debugger with trace on, ends as follows:

UTF-16:Unrecognised BM 7061, <INline 1.
at / line 0
require utf8.pm called at test4.pl line 17
main::BEGIN() called at / line 0
eval {} called at / line 0
Compilation failed in require at test4.pl line 17, <INline 1.
BEGIN failed aborted, <INline 1.
at test4.pl line 0
Debugged program terminated. Use q to quit or R to restart,

As loading utf8.pm for utf8::SWASHNEW() in utf8_heavy.pl is necessary
to compile a Unicode character class, swash_init() in utf8.c tried to
load utf8.pm if SWASHNEW is not defined, but (perhaps) wrongly assumed
it to be encoded in UTF-16, and then failed in loading it.

It seems this is fixed in Perl 5.9.4, though I'm not sure what
changed it.

Regarads,
SADAHIR Tomoyuki

Mon, 21 Aug 2006 09:53:32 +0200, demerphq wrote
8/21/06, demerphq <demerphq (AT) gmail (DOT) comwrote:
8/21/06, Andreas J. Koenig <andreas.koenig.gmwojprw (AT) franz (DOT) ak.mind.dewrote:
Sun, 20 Aug 2006 19:50:07 +0900, SADAHIR Tomoyuki <bqw10602 (AT) nifty (DOT) comsaid:

As loading utf8.pm for utf8::SWASHNEW() in utf8_heavy.pl is necessary
to compile a Unicode character class, swash_init() in utf8.c tried to
load utf8.pm if SWASHNEW is not defined, but (perhaps) wrongly assumed
it to be encoded in UTF-16, and then failed in loading it.

It seems this is fixed in Perl 5.9.4, though I'm not sure what
changed it.

I suspect

Subject: [PATCH] optimize /[x]/ to /x/.
From: demerphq <demerphq (AT) gmail (DOT) com>
Date: Sat, 20 May 2006 23:16:33 +0200
Message-Id: <@mail.gmail.com>

Yeah, that probably did it. The thing is, id expect the bug to return
if instead of

$text =~ m![t]*!;

we use

$text =~ m![tT]*!;

And a quick check shows the bug is truely gone. Which means my patch
wasnt involved

I think something other did it, too, since this issue isn't limited
to /[ ]/. As commented in utf8_heavy.pl,
cf.

instead of $text =~ m![t]*!,
(1) case-insensitive matching, like $text =~ /t/i
(2) case mapping operation for utf8, like $text = uc($text)
(3) transliteration for utf8, like $text =~ tr/t\x{100}//
(4) matching unicode properties, like $text =~ /\p{Latin}/
also trigger internal calling utf8::SWASHNEW.
And in all cases the bug seems to be gone.

Probably the change of $^H and %^H might concern it, though I must
admit I'm not sure how open.pm does with load_module().

Regards,
SADAHIR Tomoyuki