How hideous would it be if I switched back to uniform 2-part version
numbering across my modules using the "nuclear option", ie, entirely
deleting from CPAN my few (I believe lightly used) module versions
that for less than a year have borne 3-part version numbers?
Would this plan qualify as an exception to the "never" clause below?
When I asked in June about the wisdom of switching from 2-part to 3-part
version numbering, I'd already mistakenly done so for half my modules,
innocently violating the primary rule. I want to switch back because it
hurts my sense of esthetics and has caused disruption in some smokers.
-John
--- On Tue, 28 Jun 2011, David Cantrell wrote:
Do folks have thoughts on whether it's better for CPAN modules to use
2- or 3-part version numbers?
I switched recently to using "vI.J.K" version numbering with my modules
v1.2.3 isn't a number.
Nor is 1.2.3.
Therefore they are wrong.
However, because Wrongness was once thought to be Rightness, they are
supported, and using them is fine, *provided that* you are consistent
and *never* mix different types of "number" for a distribution. Once a
distribution uses proper numbers, it should *never* be changed to use
Wrongness, and if it already uses Wrongness it should *never* be changed
to use proper numbers.
(for values of "never" that are accurate enough for a brief email, I
really can't be arsed with explaining when it's OK to mix and match and
when it's not).
--
David Cantrell | Official London Perl Mongers Bad Influence
Arbeit macht Alkoholiker
