To fix a failing
#cpan DBD::Oracle
complaining about missing header files, you can specify the INCLUDE location like the following
#cpan <<< " o conf makepl_arg '-h /opt/oracle-client-11/sdk/include/' install DBD::Oracle "
With “o conf makepl_arg” we’ re just telling which parameters to pass to
#perl Makefile.PL
Obviously each module has its own ;)
After reading the previous post on PAR, you may want to unpack, modify and repack an existing PAR application.
Let’s play the game with a ficticious my-perl-webapp.bin
#mkdir /tmp/tmpdir/;
#unzip my-perl-webapp.bin -d /tmp/tmpdir/;
# cd /tmp/tmpdir
In tmpdir we’ll find:
the usual ./script/ directory with all the perl files run by the application;
the main.pl auto-generated by par;
the MANIFEST and META.yml files;
the ./lib/ directory with all the dependencies.
Now you have all the perl files and you can modify and fix whatever you want.
Check and test before rebuild!
The first time we’ll try to repack with a simple:
# pp -P -o /tmp/my-perl-webapp-1.bin script/{all .pl files but main.pl } ;
Remember that main.pl is auto-generated by PAR.
When everything is done, check if all the dependencies have been added (ex. confronting package content with unzip -t)
Repacking may require some more work. The pp command may not notify all the required dependencies you need.
You can check the MANIFEST for a list of files to add to your package.
Add to the package all the files present in the original one. You can do it with
# pp -a lib/file1.pm -a lib/file2.pm … -P -o …
To speed up things we’ll use find:
# find lib -type f -printf ” -a %p ” | \
xargs pp -P \
-o /tmp/my-perl-webapp-1.bin \
script/{all .pl files but main.pl }
-a MANIFEST \
-a META.yml
If you don’t remember how find -printf works you can check #man
Enjoy,
R.
PAR is a way to distribute packed perl applications in executable format. Something like jar files for java, but in a platform dependant way. A par file contains all the nested dependencies of a perl script and a par executor.
Shortly, let’s create a perl script
# cat < sample.pl
#!/usr/bin/perl
use Data::Dumper;
my %hash = (“a”=>1, “b”=>2);
print “hash: [“. Dumper(%a).”]”;
EOF
an make it an executable file with pp – the par packager:
# pp -o sample.bin sample.pl
Now we get a binary executable file – sample.bin – containing all the dependencies (Data::Dumper) packed as a zip file.
# ll sample.bin;
-rwxr-xr-x 1 rpolli rpolli 3.1M 2012-07-26 10:15 sample.bin
We can really unzip it (but can’t modify wit zip -u or -g) watching the anatomy of the PARchive.
# unzip -t sample.bin
script/main.pl
script/sample.pl <- our script file
MANIFEST
META.yml
…
lib/Data/Dumper.pm <- explicit dependecy
…
lib/File/Spec.pm <- implicit dependency
…
Together with the deps we have:
a main.pl file used internally by par
a MANIFEST and META.yml with packaging info
PARchives can still be create as a perl script, saving some space:
# pp -P -o sample.bin.pl sample.pl
# ll -h sample.bin.pl
-rwxr-xr-x 1 rpolli rpolli 25K 2012-07-26 10:28 sample.bin.pl
Next time we’ll see how to unpack and repack a PAR file…