Calling the question on UNIVERSAL::VERSION semantics

Using the following class file:

$ cat class1.pm
package class1;

$VERSION = 1.2.3;

1;

Here is how things stand today:


                                     class1->VERSION  $class1::VERSION

perl5.10.0 with builtin version           1.2.3              \1\2\3

perl5.12.4 with builtin version           1.2.3              \1\2\3

perl5.14.0 with builtin version           1.2.3              \1\2\3

perl5.15.3 with builtin version          \1\2\3              \1\2\3


With this class:

$ cat class2.pm
package class2;

$VERSION = '1.2.3';

1;

it obviously displays '1.2.3' for all releases.

Perl 5.10.0 shipped December 18th, 2007.  The version object code first 
shipped in 5.9.1, in April 2005, including the UNIVERSAL::VERSION 
changes.  That code itself was actually committed March 2004 after a 
nearly two year discussion and many redesigns.

I regret that I never updated the POD in UNIVERSAL.pm to reflect that 
UNIVERSAL::VERSION would return the stringified $VERSION scalar at the 
time, which would have prevented this discussion.  It is in the 
version::Internals pod

>    Replacement UNIVERSAL::VERSION
>        In addition to the version objects, this modules also replaces the core
>        UNIVERSAL::VERSION function with one that uses version objects for its
>        comparisons.  The return from this operator is always the stringified
>        form as a simple scalar (i.e. not an object), but the warning message
>        generated includes either the stringified form or the normal form,
>        depending on how it was called.

I firmly feel that maintaining the status quo where class->VERSION 
returns the stringified/normalized value is better than what was 
released in 5.15.3.  Returning the contents of the $VERSION scalar means 
that every single consumer has to figure out whether it is appropriate 
to use %vd when printing.  Both class1 and class2 have the same version; 
it seems silly to ignore that the author chose different notations to 
initialize that.

John

• Calling the question on UNIVERSAL::VERSION semantics by John Peacock