Randy,
Bottom line is that there are a lot of modules to test. While some people have the
drive to manually test each one, some don't (like me). Autrijus and crew
have put together a wonderful package which allows automated testing of modules.
I will continue to use this module, and when someone asks questions about a failing
module, I will answer to the best of my ability. I do believe FAIL grade is correct
when a module author either doesn't know about PREREQ_PM, or fails to correctly
fill theirs out completely (through forgetfulness or lazyness).
Maybe you could write some perl module which would automatically update
PREREQ_PM when doing a make dist?
Cheers,
Rob
Date sent: Sat, 11 May 2002 11:28:41 -0500 (CDT)
From: Randy Kobes <[EMAIL PROTECTED]>
To: Andreas Marcel Riechert <[EMAIL PROTECTED]>
Copies to: Autrijus Tang <[EMAIL PROTECTED]>,
Automated Perl Test Account <[EMAIL PROTECTED]>,
<[EMAIL PROTECTED]>, <[EMAIL PROTECTED]>
Subject: Re: FAIL Apache-AuthCookieDBI-1.19 i386-linux 2.4.8-
11mdkenterprise
> On 11 May 2002, Andreas Marcel Riechert wrote:
>
> > While at the moment these FAIL reports are quiet random -- meaning you
> > get a FAIL if the user has not installed the used module, but a PASS if
> > the user has installed it -- we had a discussion about exactly this topic
> > a few month ago. The common consent on this list was to send FAIL reports
> > if you catch such missing modules while runing "make test". IMHO, we
> > should continue this reporting policy.
> > For the modules author changing PREREQ_PM and uploading again is just
> > a matter of a few minutes, while it makes all our lives as users much
> > more easy. I have no idea how many of these FAIL reports -- where
> > usually I talk about PREREQ_PM -- I have sent but I never got complaints,
> > and most, if not all authors changed there Makefile.PL.
> >
> > Lets continue with these FAILS and educate the authors about PREREQ_PM.
>
> I agree about the use of PREREQ_PM being useful for users, and
> educating authors about it, but reporting a FAIL based on a
> prerequisite not being installed doesn't really contain any
> useful information on how the module fares on that particular
> platform. Another user on the exact same platform, but with the
> necessary prerequisites, may report a PASS. Or perhaps a FAIL,
> for different reasons. And then a situation might arise of
> multiple reports on the same module on essentially the same
> platform, some passing, and some failing, and having to look at
> the reports in some detail to disentangle the reasons. A casual
> browser of the cpan-testers site may not want to bother doing
> this, making the service less useful.
>
> best regards,
> randy
>
>
>
Later,
Rob
