Re: Moose Type Constraints violations to warnings?

On Tue, Jun 14, 2011 at 3:37 PM, Jesse Luehrs <[email protected]> wrote:
> On Tue, Jun 14, 2011 at 11:03:33AM +0200, Ævar Arnfjörð Bjarmason wrote:
>> On Tue, Jun 14, 2011 at 01:56, Jesse Luehrs <[email protected]> wrote:
>> > On Mon, Jun 13, 2011 at 06:45:10PM +0200, Ævar Arnfjörð Bjarmason wrote:
>> >> On Mon, Jun 13, 2011 at 18:39, Ævar Arnfjörð Bjarmason <[email protected]> wrote:
>> >> > For Mouse this monkeypatching would work:
>> >>
>> >> And this would work for Moose:
>> >>
>> >>     package Foo;
>> >>     use Moose;
>> >>     {
>> >>     package Moose::Meta::Attribute;
>> >>
>> >>     sub verify_against_type_constraint {
>> >>         my $self = shift;
>> >>         my $val  = shift;
>> >>
>> >>         return 1 if !$self->has_type_constraint;
>> >>
>> >>         my $type_constraint = $self->type_constraint;
>> >>
>> >>         $type_constraint->check($val)
>> >>             || Carp::cluck("Attribute ("
>> >>                      . $self->name
>> >>                      . ") does not pass the type constraint because: "
>> >>                      . $type_constraint->get_message($val), data => $val, @_);
>> >>     }
>> >>     }
>> >>
>> >>     has x => (is => "ro", isa => "Int");
>> >>     my $foo = Foo->new(x => "blah");
>> >>     use Data::Dumper;
>> >>     print Dumper $foo;
>> >>
>> >> And give you:
>> >>
>> >>     $ perl -Ilib /tmp/error.pl
>> >>     Subroutine RegexpRef redefined at
>> >> lib/Moose/Util/TypeConstraints/OptimizedConstraints.pm line 35.
>> >>     Subroutine verify_against_type_constraint redefined at /tmp/error.pl line 6.
>> >>     Attribute (x) does not pass the type constraint because:
>> >> Validation failed for 'Int' with value
>> >> blahdatablahinstanceFoo=HASH(0x1edadb0) at /tmp/error.pl line 14
>> >>             Moose::Meta::Attribute::verify_against_type_constraint('Moose::Meta::Attribute=HASH(0x27e0078)',
>> >> 'blah', 'instance', 'Foo=HASH(0x1edadb0)') called at
>> >> lib/Moose/Meta/Attribute.pm line 1125
>> >>             Moose::Meta::Attribute::_coerce_and_verify('Moose::Meta::Attribute=HASH(0x27e0078)',
>> >> 'blah', 'Foo=HASH(0x1edadb0)') called at lib/Moose/Meta/Attribute.pm
>> >> line 485
>> >>             Moose::Meta::Attribute::initialize_instance_slot('Moose::Meta::Attribute=HASH(0x27e0078)',
>> >> 'Moose::Meta::Instance=HASH(0x27ad5c0)', 'Foo=HASH(0x1edadb0)',
>> >> 'HASH(0x1ec8f90)') called at lib/Class/MOP/Class.pm line 575
>> >>             Class::MOP::Class::_construct_instance('Moose::Meta::Class=HASH(0x266bab0)',
>> >> 'HASH(0x1ec8f90)') called at lib/Class/MOP/Class.pm line 548
>> >>             Class::MOP::Class::new_object('Moose::Meta::Class=HASH(0x266bab0)',
>> >> 'HASH(0x1ec8f90)') called at lib/Moose/Meta/Class.pm line 252
>> >>             Moose::Meta::Class::new_object('Moose::Meta::Class=HASH(0x266bab0)',
>> >> 'HASH(0x1ec8f90)') called at lib/Moose/Object.pm line 22
>> >>             Moose::Object::new('Foo', 'x', 'blah') called at
>> >> /tmp/error.pl line 23
>> >>     $VAR1 = bless( {
>> >>                      'x' => 'blah'
>> >>                    }, 'Foo' );
>> >>
>> >
>> > For the record, please please PLEASE don't do this. Attribute traits
>> > exist for a reason.
>>
>> How would you accomplish the same thing with attribute traits?
>
> package My::Attr::Trait;
> use Moose::Role;
>
> around verify_against_type_constraint => sub {
>    my $orig = shift;
>     my $self = shift;
>    my ($val) = @_;
>
>     return 1 if !$self->has_type_constraint;
>
>     my $type_constraint = $self->type_constraint;
>
>     $type_constraint->check($val)
>         || Carp::cluck("Attribute ("
>                  . $self->name
>                  . ") does not pass the type constraint because: "
>                  . $type_constraint->get_message($val), data => $val, @_);
> };
>
> package Foo;
> use Moose;
>
> has x => (
>    traits => ['My::Attr::Trait'],
>    is     => 'ro',
>    isa    => 'Int',
> );
>
> my $foo = Foo->new(x => "blah");
> use Data::Dumper;
> print Dumper $foo;

Jesse gave me permission to expand on this code and make it a CPAN
module. I finally got around to releasing it today:

    https://metacpan.org/module/MooseX::Attribute::TypeConstraint::CustomizeFatal

I also pushed a rfc/mention-mx-attr-tc-customizefatal branch to the
Moose Git repository with a small change to the FAQ to mention that
yes, this is in fact possible, along with a blatant advert for my new
module. I hereby request review of that patch:

    $ git show -U2 --word-diff
    commit 81325e8 (HEAD, origin/rfc/mention-mx-attr-tc-customizefatal, master)
    Author: Ævar Arnfjörð Bjarmason <[email protected]>
    Date:   Wed May 2 21:42:49 2012 +0000

        Moose FAQ: Expand "Can I turn off type constraint checking?"

        Expand the "Can I turn off type constraint checking?" beyond just "no"
        to "you can do it with a MooseX extension". The shiny new extension
        happens to be mine, but also point out that if it doesn't do what you
        want it's easy to override the default behavior.

    diff --git a/lib/Moose/Manual/FAQ.pod b/lib/Moose/Manual/FAQ.pod
    index e4294d8..5a8cda0 100644
    --- a/lib/Moose/Manual/FAQ.pod
    +++ b/lib/Moose/Manual/FAQ.pod
    @@ -267,5 +267,11 @@ C<NaturalLessThanTen> constraint check.
    =head3 Can I turn off type constraint checking?

    [-Not-]{+There's no support for it in the core of Moose+} yet.
This option may
    come in a future release.

    {+Meanwhile there's a L<MooseX+}
    {+extension|MooseX::Attribute::TypeConstraint::CustomizeFatal> that+}
    {+allows you to do this on a per-attribute basis, and if it doesn't do+}
    {+what you it's easy to write one that fits your use case.+}

    =head3 My coercions stopped working with recent Moose, why did you break it?