PAR-Packer lost a line of string when packing Unicode::GCString

[saxyx]

If I have a perl file named aa.pl:

use strict;
use warnings;
use Unicode::GCString;
use utf8;

my $stringToBeMeasured = "你好，世界";
my $gcstring = Unicode::GCString->new($stringToBeMeasured);
my $columns = $gcstring->columns();

print $columns;

When I run pp -T -M Unicode::GCString --output aa.exe aa.pl to generate aa.exe, and then execute aa in the cmd, I encounter the following error message: Can't locate object method "new" via package "Unicode::GCString" at script/aa.pl line 7.

The first 10 lines of GCString.pm extracted from aa.exe are as follows:

#-*-perl-*-

package Unicode::GCString;
require 5.008;

=cut

### Pragmas:

The first 10 lines of GCString.pm from the local library are:

#-*-perl-*-

package Unicode::GCString;
require 5.008;

=encoding utf-8

=cut

### Pragmas:

Therefore, it seems that pp loses the line =encoding utf-8 when packaging Unicode::GCString.

Info:

• Win11
• Perl distribution is Strawberry Perl
• PAR Packager, version 1.062 (PAR version 1.018)

[rschupp]

PAR::Filter::PodStrip stumbles over

=encoding utf-8

=cut

and removes just the =encoding while leaving =cut. Hence perl interprets everything below =cut as POD, i.e. no perl code. As a quick workaround set the environment variable PAR_VERBATIM=1 while packing to disable use of PodStrip.

[rschupp]

BTW: pp -T -M Unicode::GCString --output aa.exe aa.pl

• don't use -T, there's no reason to ever use it
• -M Unicode::GCString is not necessary, it's automagically packed

[saxyx]

Thank you very much for your help and advice. I use -T because I don't want it to decompress and take up space every time I use the aa.exe. I learned the skill from https://stackoverflow.com/questions/34351026/how-to-reduce-par-folder-size-in-tmp.

[rschupp]

I learned the skill from https://stackoverflow.com/questions/34351026/how-to-reduce-par-folder-size-in-tmp.

That's typical stackoverflow nonsense. Don't use -T.

Here's why: create a file named Foo.pm with contents

package Foo;
sub greeting { print "hello\n" };
1;

then run

$ pp -o hello.exe -T WTF -I . -E 'use Foo; Foo::greeting();'
$ .\hello.exe
hello

so faras expected. Now change "hello" to "good bye" in Foo.pm and pack again

$ pp -o hello.exe -T WTF -I . -E 'use Foo; Foo::greeting();'
$ .\hello.exe
hello

Oops! The second incarnation of hello.exe reused the cache area of the first and didn't extract the new version of Foo.pm. If you're concerned about the stuff that accumulates in your cache area, simply blow the cache area away.

[saxyx]

Thanks again for your patient advice.

According to the documentation at https://metacpan.org/pod/pp:

-T, --tempcache Set the program unique part of the cache directory name that is used if the program is run without -C. If not set, a hash of the executable is used.

When the program is run, its contents are extracted to a temporary directory. On Unix systems, this is commonly /tmp/par-USER/cache-XXXXXXX. USER is replaced by the name of the user running the program, but "spelled" in hex. XXXXXXX is either a hash of the executable or the value passed to the -T or --tempcache switch.

The reason why

The second incarnation of hello.exe reused the cache area of the first and didn't extract the new version of Foo.pm.

is that you specified the value passed to the -T switch. It wouldn't be a problem if you didn't specify a specific value.

[rschupp]

It wouldn't be a problem if you didn't specify a specific value.

That's because -T without an argument is simply ignored. And it doesn't help "I don't want it to decompress and take up space every time I use the aa.exe" either: if you don't regenerate an executable, decompression happens on the first run, after that the cache area is reused.

[saxyx]

Thank you very much. I misunderstood earlier, now I understand what you mean.