On Tue, Nov 06, 2012 at 10:27:25AM +0900, Shmuel Fomberg wrote:
> On Tue, Nov 6, 2012 at 2:40 AM, David Cantrell wrote:
> > On Mon, Nov 05, 2012 at 04:54:00PM +0900, Shmuel Fomberg wrote:
> > > Can anyone tell me, when a cpan smoker / tester is trying to test a
> > > module but fails to install a dependency, what happens?
> > > Who should I convince to make such a failure similar to test fail?
> > No-one, because it's not a test failure, it's a missing pre-requisite.
> The module failed to install.
It didn't fail. It *correctly* didn't install.
> As a cpan author, I would have like to know
> about it.
Why? Consider that if a pre-req fails its tests, then by default the
pre-requisite's author is notified. Then he may or may not fix his bug.
If he does, great, your problem is solved. If he doesn't, well, I don't
see how harrassing you (and every other author who depends on his code)
is going to help.
But in any case, authors get Really Pissed Off if they start getting
failure reports because of missing dependencies, so the CPAN-testing
software tries hard not to generate them. If you *do* want them
generated, then you need to:
* come up with a reliable way of differentiating the specific types of
pre-requisite failure that you're interested in from all others;
* convince cpan-testers-discuss that this would be useful to others
and not just you;
* convince people that it's worth their while updating the testing
software and the stuff that runs the cpantesters website
that last one is the difficult bit. People lack tuits and there are far
more important tasks waiting in the to-do list.
However, if you still want to do this, then I suggest that what you want
is a new type of report alongside PASS/FAIL/NA/UNKNOWN - PREREQFAIL.
Those reports would, by default, *not* be forwarded from the metabase to
authors.
> I don't think that the users care why it failed to install.
> So how would I fix the dependency problem if I don't even know about it?
By testing your code before releasing it? When I release a new
distribution, or make a significant change to an old dist, I test it in
half a dozen different builds of perl with no pre-reqs installed. And I
try to remember to *always* specify what version of a module I depend
on. For the few of my dists that I care most about, I re-test them
every so often to make sure that CPAN didn't break them.
> > This is why <http://deps.cpantesters.org/> and
> > <http://analysis.cpantesters.org/> exist.
> And why would an author go visit these sites if he doesn't know that there
> is a problem?
I wrote deps.cpantesters.org because I wanted to be able have a look at
the quality of any pre-requisites before using them.
--
David Cantrell | top google result for "internet beard fetish club"
comparative and superlative explained:
<Huhn> worse, worser, worsest, worsted, wasted
