On Tue, Jun 21, 2011 at 07:41:42AM -0700, Jeff Lavallee wrote:
> If you actually install modules, how do you identify modules that don't
> properly declare their dependencies?
They'll fail their tests if they try to use a module that isn't
installed.
My smokers run through about a day's worth of CPAN uploads at a time,
installing everything (including dependencies) as they go, before
cleaning everything out and starting the next batch with a clean build
of perl.
Yes, there's a chance that I might send a PASS report for something
which doesn't properly declare its dependencies but they just happen to
already be installed because something else depended on them. I
consider this to be an acceptable source of a small number of unimportant
errors, given how reliable and simple it makes things.
I could clean everything out for each distribution that I test, but that
would be intolerably slow if one day someone uploaded several modules
all of which depend on big fat bloaters like MooseX::*, which I would
have to re-test and install every time. In fact, I used to do that, and
some of my slower machines just couldn't keep up with CPAN uploads.
--
David Cantrell | even more awesome than a panda-fur coat
fdisk format reinstall, doo-dah, doo-dah;
fdisk format reinstall, it's the Windows way
