$ cat a.pl
use strict;
use warnings;
use feature qw( say );

use Text::CSV qw( );
use Encode qw( encode_utf8 );

say Text::CSV->VERSION;
say Text::CSV->is_xs() ? 'xs' : 'pp';
say Text::CSV->version;

my $csv = Text::CSV->new ({
   allow_whitespace => 1,
   binary           => 1,
   sep_char         => '|',
}) or die;

my $csv_file = encode_utf8(<<'__EOI__');
foo|bar|baz
dog|cat|cow
__EOI__

open(my $csv_fh, "<:encoding(utf8)", \$csv_file) or die;

say for $csv->column_names($csv->getline($csv_fh));

$ perl a.pl
1.06
xs
0.52
foo
bar
baz
[download]

And with a locally installed Perl 5.12.1 on the same machine:

$ perl a.pl
1.20
xs
0.74
foo
bar
baz
[download]

Yes, they updated to 5.10.0 over the summer - broke my code on there only insofaras I had to recompile the CPAN modules I was using due to the binary incompatibilities. However, to answer the question, then:

$ perl -MText::CSV_XS -le 'print $Text::CSV_XS::VERSION'
0.52
[download]

So I upgraded Text::CSV (and Text::CSV_XS)

$ perl -MText::CSV -e 'print "$Text::CSV::VERSION\n"'
1.06
$ eval $(perl -I$HOME/perl5/lib/perl5 -Mlocal::lib)
$ perl -MText::CSV -e 'print "$Text::CSV::VERSION\n"'
1.20
[download]

And I'm getting somewhere, now I'm seeing a new error:

 EHR - getline_hr () called before column_names () at myscript.pl line
+ 676
[download]

however I believe I called column_names:

# declare column names for getline_hr
$csv->column_names($csv->getline($csv_fh));

# Build Array of HashRefs 
my @buttons;
while (my $row = $csv->getline_hr($csv_fh))
    {
    push(@buttons, $row) unless ($. == 1);
    }
[download]

OK I think I see the problem. getline IS called before column_names:

$csv->column_names($csv->getline($csv_fh));
[download]

Now to fix it...

Make sure that getline was successful

As of Text-CSV_XS version 0.71, you can pass the auto_diag => 1 attribute to the new () constructor (actually already in version 0.66, but in 0.71 it was fully functional. For Text::CSV these were versions 1.14 and 1.17), and all diagnostics will be automatic.


Enjoy, Have FUN! H.Merijn

broke my code on there only insofaras I had to recompile the CPAN modules I was using due to the binary incompatibilities

That made me wonder why INSTALL_BASE is recommended over PREFIX. Using the latter, XS modules would go missing rather than become invalid, and you could actually have multiple perl installs using the same lib dir (which is good since they all use the same ENV var to locate the lib dir).

I just use perlbrew now.