t('name')

This subroutine lets you access any types you have declared so far, as well as any types you imported from another type library.

If you pass an unknown name, it throws an exception.

declare(...)

This subroutine declares a named type. The first argument is the type name, followed by a set of key/value parameters:

• parent => $type

  The parent should be another type object. Specifically, it can be anything which does the Specio::Constraint::Role::Interface role. The parent can be a named or anonymous type.

• where => sub { ... }

  This is a subroutine which defines the type constraint. It will be passed a single argument, the value to check, and it should return true or false to indicate whether or not the value is valid for the type.

  This parameter is mutually exclusive with the "inline" parameter.

• inline => sub { ... }

  This is a subroutine that is called to generate inline code to validate the type. Inlining can be much faster than simply providing a subroutine with the "where" parameter, but is often more complicated to get right.

  The inline generator is called as a method on the type with one argument. This argument is a string containing the variable name to use in the generated code. Typically this is something like '$_[0]' or '$value'.

  The inline generator subroutine should return a string of code representing a single term, and it should not be terminated with a semicolon. This allows the inlined code to be safely included in an "if" statement, for example. You can use "do { }" blocks and ternaries to get everything into one term. Do not assign to the variable you are testing. This single term should evaluate to true or false.

  The inline generator is expected to include code to implement both the current type and all its parents. Typically, the easiest way to do this is to write a subroutine something like this:

    sub {
        my $self = shift;
        my $var  = shift;
        return $self->parent->inline_check($var)
            . ' and more checking code goes here';
    }
      

  Or, more concisely:

    sub { $_[0]->parent->inline_check( $_[1] ) . 'more code that checks $_[1]' }
      

  The "inline" parameter is mutually exclusive with the "where" parameter.

• message_generator => sub { ... }

  A subroutine to generate an error message when the type check fails. The default message says something like "Validation failed for type named Int declared in package Specio::Library::Builtins (.../Specio/blib/lib/Specio/Library/Builtins.pm) at line 147 in sub named (eval) with value 1.1".

  You can override this to provide something more specific about the way the type failed.

  The subroutine you provide will be called as a method on the type with two arguments. The first is the description of the type (the bit in the message above that starts with "type named Int ..." and ends with "... in sub named (eval)". This description says what the thing is and where it was defined.

  The second argument is the value that failed the type check, after any coercions that might have been applied.

anon(...)

This subroutine declares an anonymous type. It is identical to "declare" except that it expects a list of key/value parameters without a type name as the first parameter.

coerce(...)

This declares a coercion from one type to another. The first argument should be an object which does the Specio::Constraint::Role::Interface role. This can be either a named or anonymous type. This type is the type that the coercion is to.

The remaining arguments are key/value parameters:

• from => $type

  This must be an object which does the Specio::Constraint::Role::Interface role. This is type that we are coercing from. Again, this can be either a named or anonymous type.

• using => sub { ... }

  This is a subroutine which defines the type coercion. It will be passed a single argument, the value to coerce. It should return a new value of the type this coercion is to.

  This parameter is mutually exclusive with the "inline" parameter.

• inline => sub { ... }

  This is a subroutine that is called to generate inline code to perform the coercion.

  The inline generator is called as a method on the type with one argument. This argument is a string containing the variable name to use in the generated code. Typically this is something like '$_[0]' or '$value'.

  The inline generator subroutine should return a string of code representing a single term, and it should not be terminated with a semicolon. This allows the inlined code to be safely included in an "if" statement, for example. You can use "do { }" blocks and ternaries to get everything into one term. This single term should evaluate to the new value.

any_isa_type, object_isa_type

The "any_isa_type" helper creates a type which accepts a class name or object of the given class. The "object_isa_type" helper creates a type which only accepts an object of the given class.

These subroutines take a type name as the first argument. The remaining arguments are key/value pairs. Currently this is just the "class" key, which should be a class name. This is the class that the type requires.

The type name argument can be omitted to create an anonymous type.

You can also pass just a single argument, in which case that will be used as both the type's name and the class for the constraint to check.

any_does_type, object_does_type

The "any_does_type" helper creates a type which accepts a class name or object which does the given role. The "object_does_type" helper creates a type which only accepts an object which does the given role.

These subroutines take a type name as the first argument. The remaining arguments are key/value pairs. Currently this is just the "role" key, which should be a role name. This is the class that the type requires.

This should just work (I hope) with roles created by Moose, Mouse, and Moo (using Role::Tiny).

The type name argument can be omitted to create an anonymous type.

You can also pass just a single argument, in which case that will be used as both the type's name and the role for the constraint to check.

any_can_type, object_can_type

The "any_can_type" helper creates a type which accepts a class name or object with the given methods. The "object_can_type" helper creates a type which only accepts an object with the given methods.

These subroutines take a type name as the first argument. The remaining arguments are key/value pairs. Currently this is just the "methods" key, which can be either a string or array reference of strings. These strings are the required methods for the type.

The type name argument can be omitted to create an anonymous type.

enum

This creates a type which accepts a string matching a given list of acceptable values.

The first argument is the type name. The remaining arguments are key/value pairs. Currently this is just the "values" key. This should an array reference of acceptable string values.

The type name argument can be omitted to create an anonymous type.

intersection

This creates a type which is the intersection of two or more other types. A union only accepts values which match all of its underlying types.

The first argument is the type name. The remaining arguments are key/value pairs. Currently this is just the "of" key. This should an array reference of types.

The type name argument can be omitted to create an anonymous type.

union

This creates a type which is the union of two or more other types. A union accepts any of its underlying types.

The first argument is the type name. The remaining arguments are key/value pairs. Currently this is just the "of" key. This should an array reference of types.

The type name argument can be omitted to create an anonymous type.