I'm trying to set up a common (checked into version-control) copy of 
Ruby that all our users can access using Windows or Linux clients.
One benefit of the version control mechanism is that we have built-in 
update policies for keeping users local directories in sync with a 
central store.
This is ideal for the Ruby installation, as it and our internal scripts 
are updated fairly regularly.

The only problem is that Ruby seems to play by different rules under 
Linux and Windows.  Under Windows, Ruby's standard $: directories 
reference lib/ruby/1.8/... and lib/ruby/site_ruby/.... underneath the 
Ruby install directory.
Under Linux, Ruby is in /usr/bin, but $: references /usr/lib/ruby and 
/usr/lib/site_ruby.  How can I have a single directory tree containing 
Ruby for Linux *and* Ruby for Windows?  I would think it should easily 
work (.so's are already broken out into platform specific directories), 
but I'm not sure what to change with the Linux version to make it act 
like the Windows version.

So, how do I change what the Linux install of Ruby uses for its default 
$: search path?

Thanks,
Patrick Bennett