On Thu, 23 Jun 2005 09:26:33 +0000
"Johnny C. Lam" <jlam@netbsd.org> wrote:

 
> To summarize, I think BSDPAN is the wrong way to solve the problem
> you're describing.
> 
> 	Cheers,
> 
> 	-- Johnny Lam <jlam@NetBSD.org>
> 

FreeBSD does it. Thats a compelling story in userspace (less so in
kernel, but divergence is not a goal in itself)

I have experienced using this system. To date, it hasn't stuffed up for
me, and had the benefit of giving a lightweight method of tracking CPAN
cleanly enough.

CPAN is a mess. a huge mess. specific versions disappear. the namespace
is fluid. the CPAN tool itself is a mess. each user carries the .cpan
install history they did, no common DB.

This BSDPAN fix puts a single common checkable install trace for code
in, if it uses MakeMaker and its associated Makefile.PL

If you can't come at a globally applied solution, could it be an
optional package to apply over perl so we can choose to do it for
ourselves?

-george