All,
This email refers to an RT ticket I've filed that I think warrants
some discussion
(https://rt.cpan.org/Public/Bug/Display.html?id=44873).
I've created a rudimentary ORM that maps attributes in a class to
columns in a table and maps an instance of a class to a table row.
This has proven
On Mon, Apr 13, 2009 at 12:38:08PM -0500, Dan Harbin wrote:
I'd like for the following validation in the Class::MOP::Attribute
constructor to be removed:
(defined $name $name)
|| confess You must provide a name for the attribute;
Is there any reason Class::MOP::Attribute should
Dan Harbin wrote:
For instance, I want a column named 0 (the number zero). While that
is valid as a mysql column, that is invalid as an attribute name. The
only reason I can fathom that 0 is prohibited from being an attribute
name is because $obj-0 is invalid perl, but even so you can still
I've attached to the RT ticket some tests that Stevan was suggesting:
https://rt.cpan.org/Ticket/Attachment/590690/299095/test.pl
I'm not sure if it's comprehensive, but at least it's a start. It
uncovered the same (defined $name $name) check in:
* Class::MOP::Class::get_attribute
*
I added a note in the Moose::Meta::Role documentation yesterday to the effect
that applying a role did not do any checks to make sure that the target had not
already consumed the role you were attempting to apply.
I always find myself writing apply role unless does role.
Would anyone object to
