Re: Sysadmin questions on efficient installation of new Perl versions

This would not even be quite as big a problem except that 1) it means that some of our applications are down until we can get every module re-installed, and 2) It is very difficult to keep track of every module that is installed so we can make sure that we get them all re-installed.

For (1) I tend to build perl in it's own tree, install modules I need, and then symlink to it. That way the down time is almost zero when upgrading to a new version (just move the symlink), and if you run into a problem during the build it's no big deal. Also, it's easy to move Perl to machines that have a similar architecture and OS version, which is hard to do if it is installed somewhere like /usr/local where everything is mixed together.

For (2) I punt. I can't use CPAN or CPANPLUS, for various reasons, and so I tend to keep the original source dirs for these around as well and rebuild them -- not much fun.

Comment on Re: Sysadmin questions on efficient installation of new Perl versions