Hi folks, I'm revisiting a problem from a couple years back in hopes that a newer, better solution has or can be created:
I want to automate the process of building a collection of CPAN modules into a non-standard directory (so that I can distribute the built collection as part of another piece software) and I want the installation process to ignore any already installed, non-core modules. For example, my list of modules to install might include Params::Validate and I want to install using a prefix of /tmp/INSTALL_DIR Given that Params::Validate depends upon: Attribute::Handlers: 0.79 Scalar::Util: 1.10 Test::More: 0 and that on my build machine: 'extraslib' => '/System/Library/Perl/Extras/5.10.0' 'privlib' => '/System/Library/Perl/5.10.0' and that on my build machine Attribute::Handlers and Test::More are installed in privlib but Scalar::Util is installed in updateslib (/Library/Perl/Updates/5.10.0/darwin-thread-multi-2level/Scalar/Util.pm) I want my automated build to install in /tmp/INSTALL_DIR not only Params::Validate but also Scalar::Util because my automated build process should ignore the Scalar::Util that is instaled in updateslib. In the past I have achieved this by having a complete hierarchy of the "baseline Perl" modules - the ones to ignore, and by replacing two CPAN functions at run-time: CPAN::Module::inst_version CPAN::Module::available_file I replace them with functions that use the stored hierarchy of the "baseline Perl" modules instead of the currently running Perl configuration. This works, but is messy and not very elegant. Does anyone have a better solution? -Matisse
