Thanks for the concrete example!

I'm thinking though it might be easier to link the codebase lib directory into the build directory rather than the other way around, e.g.

#error checking, use... omitted
#may have typos - read as perl-ish pseudocode

my @buildirs = ("foo1", "foo2");
my $repo="myrepo"; 
my $startdir=File::Spec::curdir();

foreach my $buildme (@buildirs) {
   chdir($buildme); 
   symlink($repo, "lib");
   system("perl Build.PL");
   unlink("lib");
}
chdir($startdir);
[download]

The advantage of this approach is that I don't have to have to make assumptions about the support files used by the build process, e.g. is there a MANIFEST.SKIP? is there a special subclass of Module::Build that needs other kinds of support files?, etc, etc.

The main disadvantage of any linking approach is that it isn't well behaved on MS-Win platforms. Although I do most of my development on Linux, there are situations when I need to work on MS machines. Cygwin/Vista has a particularly nasty bug where un-linking a symbolic link defined in Vista via anything using cygwin libraries not only removes the link, but also the underlying file or directory!

beth