Steve Peters <[EMAIL PROTECTED]> writes:
> Unfortunately, the new version of podlators causes test failures in
> lib/Pod/t/Usage.t.(*)
The test suite is capturing output by tying the output file handle to a
class that stores the output in a string, but that class doesn't implement
close. Pod::Man and Pod::Text in 2.0.1 close the filehandle in
parse_from_file to force a flush of output since Pod::Simple otherwise
doesn't flush. The more I think about it, that's the wrong thing to do,
so I've changed the code to just flush the file handle instead.
Incidentally, can anyone tell me why I have to do:
my $fh = $self->output_fh ();
my $oldfh = select $fh;
my $oldflush = $|;
$| = 1;
print $fh '';
$| = $oldflush;
select $oldfh;
instead of the obvious:
my $fh = $self->output_fh ();
$fh->flush ();
which the IO::Handle documentation seems to indicate is supposed to work?
If I do the latter, I just get "Can't locate object method "flush" via
package "IO::Handle" at ../blib/lib/Pod/Text.pm line 594." This is with
Perl 5.8.7.
> I also must have forgotten to send you updates to t/man.t. This change
> was to TODO various tests that were not Unicode friendly. The patch for
> this is included below against podlators-2.0.
Pod::Man supports Unicode, so it took me a little bit to figure out what
was really wrong here. The right fix is:
--- t/man.t 20 Jan 2006 21:20:58 -0000 1.6
+++ t/man.t 25 Jan 2006 23:14:24 -0000
@@ -34,6 +34,12 @@ my $n = 2;
while (<DATA>) {
next until $_ eq "###\n";
open (TMP, '> tmp.pod') or die "Cannot create tmp.pod: $!\n";
+
+ # We have a test in ISO 8859-1 encoding. Make sure that nothing strange
+ # happens if Perl thinks the world is Unicode. Wrap this in eval so that
+ # older versions of Perl don't croak.
+ eval { binmode (\*TMP, ':encoding(iso-8859-1)') };
+
while (<DATA>) {
last if $_ eq "###\n";
print TMP $_;
@@ -139,6 +145,8 @@ Also not a bullet.
###
###
+=encoding iso-8859-1
+
=head1 ACCENTS
Beyoncé! Beyoncé! Beyoncé!!
Two things are needed: first, the temporary file to which the POD is
written has to not have a UTF-8 layer or otherwise the input is
automatically translated by Perl into UTF-8 as near as I can tell, and
second, there was a missing =encoding to tell Pod::Simple what the
encoding of the input is.
With this fix, everything looks good in Perl 5.8.7 and in 5.6.1. A new
release is coming in a few minutes.
--
Russ Allbery ([EMAIL PROTECTED]) <http://www.eyrie.org/~eagle/>
