Re: testing CPAN-ability without CPAN?

>>>>> On Tue, 20 Dec 2005 08:43:20 -0800, Tyler MacDonald <[email protected]> said:

  > Tyler MacDonald <[email protected]> wrote:
 >> > Aside from the suggestion from Thomas Klausner, you could subclass 
 >> > CPAN::Mini to do just that. Add Module::Dependency and you should get 
 >> > the system up and running rather quickly. I fact, it would probably be a 
 >> > really good idea to make the result availlable on CPAN as 
 >> > CPAN::Mini::Something. (Ask the modules list for a good namespace choice 
 >> > or #perl or Ricardo Signes, the author of CPAN::Mini.)

  > 	I should mention also that what I'm ultimately after is something
  > like "cpans_ok" and "depends_ok" for unit tests. For the "depends_ok"
  > action, I'd want to test that when the dependancies are present, you get a
  > makefile (or buildfile), and when the dependancies are _not_ present, CPAN
  > picks up on that (which I'm guessing would involve overloading
  > CPAN::follow_prereqs).

  > 	Of course, if you're to the point of unit testing a module, you're
  > either developing it, or your own CPAN has probably already followed_prereqs
  > for you... so there's got to be a way to "fake" the prereqs not being there
  > as well. *Ideally* (since Makefile.PL/Build.PL behaviour may differ if a
  > module is or isn't there), I'd like to fool perl into thinking that those
  > files aren't in @INC when they actually are. The best I've come up with for
  > solving that, is to create another directory tree with empty files with the
  > same names as the modules (tmp/Foo/Bar.pm, etc) and popping that onto
  > @INC... is there a better way to do this that doesn't involve creating a
  > bunch of stub files?

I'm not sure I can follow you, but did you consider

 @INC = ();

?

-- 
andreas