Randy Kobes <[EMAIL PROTECTED]> writes:
> or other reasons an author may have ... not wanting to belabour
> the point, but one problem I have with this policy is consistency
> - on the same system, the grade would depend upon the order used
> to test a set of packages.
While I still favour sending out FAIL's a also see the consistency
problem. It's a kind of random wether the author gets a FAIL or PASS.
>From my experience I can say, that most authors who get such a FAIL
report will upload a fixed version in a very short time. So in practice
our consistency problem usually goes away for most "newest version"
modules in a short time. In theory and for a few real modules the problem
still exists.
Basically there remain two questions:
(1) Can we check reliable wether all prerequisites are set
in PREREQ_PM?
(Some routine which does not care wether this PREREQ's are
locally installed or not.) ... If we find a way to do this,
all our Problems are gone and we can send FAILS without problems)
(2) Which does more harm ... Sending FAILs which may be not
consistent or sending a PASS which does not wake up the author ergo
loosing some chance to improve KWALITEE on CPAN?
Andreas
