On Wednesday 26 September 2007 7:08 pm, David Golden wrote: > On 9/26/07, Graham TerMarsch <[EMAIL PROTECTED]> wrote: > > What I'd like, though, is to at least stem/slow the number of CPAN > > testers failures or at least get them turned into something more like "no > > idea; build environment not suitable for this module". > > At least with CPAN, I think that if you just exit the Makefile.PL > without creating a Makefile, then CPAN::Reporter won't FAIL it, but > CPAN notices that no Makefile was created and won't continue. But you > have to exit (exit value 0), not die. > > I don't know if CPANPLUS follows the same logic.
Ok... THATS a useful answer. :) So... if I set up my Makefile.PL to check for the presence of any modules that I require for configuration (e.g. "Apache::Test") and then simply "exit 0" -before- the Makefile is written, it'll abort the build/test and -NOT- record a failure? -=SCHWEEET=- That's what I'm looking for. That'll give me a chance to spit a warning out to the user to let them know that they have to install these modules -before- running "perl Makefile.PL". It'll also mean that automated test reporters like CPANPLUS will stop filing "failure" reports when its really the test environment that's lacking. So... would anyone care to fill me in on why we don't have CPAN(PLUS) et al. re-running "perl Makefile.PL" after dependencies are installed? I'm thinking it might cause problems for users who are installing packages that prompt them for input during that run, but also think that it'd be useful for cases like this where Makefile.PL has to change its configuration based on whether or not the dependencies are installed. -- Graham TerMarsch Howling Frog Internet Development, Inc.
