The London Perl and Raku Workshop takes place on 26th Oct 2024. If your company depends on Perl, please consider sponsoring and/or attending.

NAME

Object::Pad - a simple syntax for lexical field-based objects

SYNOPSIS

On perl version 5.26 onwards:

   use v5.26;
   use Object::Pad;

   class Point {
      field $x :param = 0;
      field $y :param = 0;

      method move ($dX, $dY) {
         $x += $dX;
         $y += $dY;
      }

      method describe () {
         print "A point at ($x, $y)\n";
      }
   }

   Point->new(x => 5, y => 10)->describe;

Or, for older perls that lack signatures:

   use Object::Pad;

   class Point {
      field $x :param = 0;
      field $y :param = 0;

      method move {
         my ($dX, $dY) = @_;
         $x += $dX;
         $y += $dY;
      }

      method describe {
         print "A point at ($x, $y)\n";
      }
   }

   Point->new(x => 5, y => 10)->describe;

DESCRIPTION

This module provides a simple syntax for creating object classes, which uses private variables that look like lexicals as object member fields.

While most of this module has evolved into a stable state in practice, parts remain experimental because the design is still evolving, and many features and ideas have yet to implemented. I don't yet guarantee I won't have to change existing details in order to continue its development. Feel free to try it out in experimental or newly-developed code, but don't complain if a later version is incompatible with your current code and you'll have to change it.

That all said, please do get in contact if you find the module overall useful. The more feedback you provide in terms of what features you are using, what you find works, and what doesn't, will help the ongoing development and hopefully eventual stability of the design. See the "FEEDBACK" section.

Experimental Features

Since version 0.63.

Some of the features of this module are currently marked as experimental. They will provoke warnings in the experimental category, unless silenced.

You can silence this with no warnings 'experimental' but then that will silence every experimental warning, which may hide others unintentionally. For a more fine-grained approach you can instead use the import line for this module to only silence the module's warnings selectively:

   use Object::Pad ':experimental(mop)';

   use Object::Pad ':experimental(custom_field_attr)';

   use Object::Pad ':experimental(composed_adjust)';

   use Object::Pad ':experimental(inherit_field)';

   use Object::Pad ':experimental';  # all of the above

Since version 0.64.

Multiple experimental features can be enabled at once by giving multiple names in the parens, separated by spaces:

   use Object::Pad ':experimental(mop custom_field_attr)';

Automatic Construction

Classes are automatically provided with a constructor method, called new, which helps create the object instances. This may respond to passed arguments, automatically assigning values of fields, and invoking other blocks of code provided by the class. It proceeds in the following stages:

The BUILDARGS phase

If the class provides a BUILDARGS class method, that is used to mangle the list of arguments before the BUILD blocks are called. Note this must be a class method not an instance method (and so implemented using sub). It should perform any SUPER chaining as may be required.

   @args = $class->BUILDARGS( @_ )

Field assignment

If any field in the class has the :param attribute, then the constructor will expect to receive its argmuents in an even-sized list of name/value pairs. This applies even to fields inherited from the parent class or applied roles. It is therefore a good idea to shape the parameters to the constructor in this way in roles, and in classes if you intend your class to be extended.

The constructor will also check for required parameters (these are all the parameters for fields that do not have default initialisation expressions). If any of these are missing an exception is thrown.

The BUILD phase

As part of the construction process, the BUILD block of every component class will be invoked, passing in the list of arguments the constructor was invoked with. Each class should perform its required setup behaviour, but does not need to chain to the SUPER class first; this is handled automatically.

The ADJUST phase

Next, the ADJUST block of every component class is invoked. This happens after the fields are assigned their initial values and the BUILD blocks have been run.

The strict-checking phase

Finally, before the object is returned, if the ":strict(params)" class attribute is present, then the constructor will throw an exception if there are any remaining named arguments left over after assigning them to fields as per :param declarations, and running any ADJUST blocks.

KEYWORDS

class

   class Name :ATTRS... {
      ...
   }

   class Name :ATTRS...;

Behaves similarly to the package keyword, but provides a package that defines a new class. Such a class provides an automatic constructor method called new.

As with package, an optional block may be provided. If so, the contents of that block define the new class and the preceding package continues afterwards. If not, it sets the class as the package context of following keywords and definitions.

As with package, an optional version declaration may be given. If so, this sets the value of the package's $VERSION variable.

   class Name VERSION { ... }

   class Name VERSION;

An optional list of attributes may be supplied in similar syntax as for subs or lexical variables. (These are annotations about the class itself; the concept should not be confused with per-object-instance data, which here is called "fields").

Whitespace is permitted within the value and is automatically trimmed, but as standard Perl parsing rules, no space is permitted between the attribute's name and the open parenthesis of its value:

   :attr( value here )     # is permitted
   :attr (value here)      # not permitted

The following class attributes are supported:

:isa

   :isa(CLASS)

   :isa(CLASS CLASSVER)

Since version 0.57.

Declares a superclass that this class extends. At most one superclass is supported.

If the package providing the superclass does not exist, an attempt is made to load it by code equivalent to

   require CLASS;

and thus it must either already exist, or be locatable via the usual @INC mechanisms.

The superclass may or may not itself be implemented by Object::Pad, but if it is not then see "SUBCLASSING CLASSIC PERL CLASSES" for further detail on the semantics of how this operates.

An optional version check can also be supplied; it performs the equivalent of

   BaseClass->VERSION( $ver )

:does

   :does(ROLE)

   :does(ROLE ROLEVER)

Since version 0.57.

Composes a role into the class; optionally requiring a version check on the role package.

Multiple roles can be composed by using multiple :does attributes, one per role.

The package will be loaded in a similar way to how the ":isa" attribute is handled.

:repr(TYPE)

Sets the representation type for instances of this class. Must be one of the following values:

   :repr(native)

The native representation. This is an opaque representation type whose contents are not specified. It only works for classes whose entire inheritance hierarchy is built only from classes based on Object::Pad.

   :repr(HASH)

The representation will be a blessed hash reference. The instance data will be stored in an array referenced by a key called Object::Pad/slots, which is fairly unlikely to clash with existing storage on the instance. No other keys will be used; they are available for implementions and subclasses to use. The exact format of the value stored here is not specified and may change between module versions, though it can be relied on to be well-behaved as some kind of perl data structure for purposes of modules like Data::Dumper or serialisation into things like YAML or JSON.

   :repr(keys)

Since version 0.803.

The representation will be a blessed hash reference. The instance data will be stored in individual keys of the hash, named after the class and the field variable name, separated by a / symbol. Objects in this representation should behave predictably with data printing modules like Data::Dumper or serialisation via YAML or JSON.

These two hash-based representation types may be useful when converting existing classes into using Object::Pad where there may be existing subclasses of it that presume a blessed hash for their own use.

   :repr(magic)

The representation will use MAGIC to apply the instance data in a way that is invisible at the Perl level, and shouldn't get in the way of other things the instance is doing even in XS modules.

This representation type is the only one that will work for subclassing existing classes that do not use blessed hashes.

   :repr(pvobj)

Since version 0.804.

The representation will be the SVt_PVOBJ type newly added to Perl, which offers more efficient storage for object instances. This is only available on Perl version 5.38.0 onwards.

This is also newly-added and may not be fully tested and reliable yet. Once it has more real-world testing and has proven reliable it may become the default instance representation on versions of Perl where it is available.

   :repr(autoselect), :repr(default)

Since version 0.23.

This representation will select one of the representations above depending on what is best for the situation. Classes not derived from a non-Object::Pad base class will pick native, and classes derived from non-Object::Pad bases will pick either the HASH or magic forms depending on whether the instance is a blessed hash reference or some other kind.

This achieves the best combination of DWIM while still allowing the common forms of hash reference to be inspected by Data::Dumper, etc. This is the default representation type, and does not have to be specifically requested.

:strict(params)

Since version 0.43.

Can only be applied to classes that contain no BUILD blocks. If set, then the constructor will complain about any unrecognised named arguments passed to it (i.e. names that do not correspond to the :param of any defined field and left unconsumed by any ADJUST block).

Since BUILD blocks can inspect the arguments arbitrarily, the presence of any such block means the constructor cannot determine which named arguments are not recognised.

This attribute is a temporary stepping-stone for compatibility with existing code. It is recommended to enable this whenever possible, as a later version of this module will likely perform this behaviour unconditionally whenever no BUILD blocks are present.

class (anon)

   my $class = class :ATTRS... { ... };

Since version 0.809.

If a class keyword is not followed by a package name, it creates an anonymous class expression. This is an expression that yields a value suitable to use as a constructor invocant for creating instances of that class, without specifying what its package name will actually be.

This is useful for creating small one-off instances inline in expressions, such as in unit tests. Since it still accepts the usual attributes and inner body statements, it can be useful for creating one-off instances of roles, with required methods being applied.

   my $testobj = (class {
      apply Role::Under::Test;
      method required { return "a useful value"; }
   })->new;

Due to limitations on how classes work in Perl, anonymous classes are still backed by long-lived named classes in the global symbol table, unlike true anonymous functions which can go out of scope and be reclaimed once no references to them remain in existence. This means that anonymous classes will retain references to any variables captured within them, even if the class expression itself goes out of scope and any instances created by it no longer remain.

role

   role Name :ATTRS... {
      ...
   }

   role Name :ATTRS...;

Since version 0.32.

Similar to class, but provides a package that defines a new role. A role acts similar to a class in some respects, and differently in others.

Like a class, a role can have a version, and named methods.

   role Name VERSION {
      method a { ... }
      method b { ... }
   }

A role does not provide a constructor, and instances cannot directly be constructed. A role cannot extend a class.

A role can declare that it requires methods of given names from any class that implements the role.

   role Name {
      requires METHOD;
   }

A role can provide instance fields. These are visible to any ADJUST blocks or methods provided by that role.

Since version 0.33.

   role Name {
      field $f;

      ADJUST { $f = "a value"; }

      method f { return $f; }
   }

Since version 0.57 a role can declare that it provides another role:

   role Name :does(OTHERROLE) { ... }
   role Name :does(OTHERROLE OTHERVER) { ... }

This will include all of the methods from the included role. Effectively this means that applying the "outer" role to a class will imply applying the other role as well.

The following role attributes are supported:

:compat(invokable)

Since version 0.35.

Enables a form of backward-compatibility behaviour useful for gradually upgrading existing code from classical Perl inheritance or mixins into using roles.

Normally, methods of a role cannot be directly invoked and the role must be applied to an Object::Pad-based class in order to be used. This however presents a problem when gradually upgrading existing code that already uses techniques like roles, multiple inheritance or mixins when that code may be split across multiple distributions, or for some other reason cannot be upgraded all at once. Methods within a role that has the :compat(invokable) attribute applied to it may be directly invoked on any object instance. This allows the creation of a role that can still provide code for existing classes written in classical Perl that has not yet been rewritten to use Object::Pad.

The tradeoff is that a :compat(invokable) role may not create field data using the "field" keyword. Whatever behaviours the role wishes to perform must be provided only by calling other methods on $self, or perhaps by making assumptions about the representation type of instances.

It should be stressed again: This option is only intended for gradual upgrade of existing classical Perl code into using Object::Pad. When all existing code is using Object::Pad then this attribute can be removed from the role.

inherit

   inherit Classname;
   inherit Classname VER;

   inherit Classname LIST...;
   inherit Classname VER LIST...;

Declares a superclass that this class extends. At most one superclass is supported. If present, this declaration must come before any methods or fields are declared, or any roles applied. (Other compile-time declarations such as use statements that import utility functions or other behaviours may be permitted before this, however, provided that they do not interact with the class structure in any way).

This is a newer form of the :isa attribute intended to be more flexible if import arguments or other features are added at a later time.

If the package providing the superclass does not exist, an attempt is made to load it by code equivalent to

   require Classname;

and thus it must either already exist, or be locatable via the usual @INC mechanisms.

An optional version check can also be supplied; it performs the equivalent of

   Classname->VERSION( $ver )

Experimentally since version 0.807, an optional list of arguments can also be provided, in similar syntax to those in a use statement. Currently this list of arguments must be names of fields to be inherited. Only fields in the base class that are annotated with the :inheritable attribute may be inherited. Once a field is inherited, methods and other expressions in the class body can use that field identically to any fields defined by that class itself.

   class Class1 {
      field $x :inheritable = 123;
   }

   class Class2 {
      inherit Class1 '$x';
      field $y = 456;
      method describe { say "Class2(x=$x,y=$y)" }
   }

   Class2->new->describe;

apply

   apply Rolename;

   apply Rolename VER;

Since version 0.807.

Composes a role into the class; optionally requiring a version check on the role package. This is a newer form of the :does attribute intended to be more flexible if import arguments or other features are added at a later time.

Multiple roles can be composed by using multiple :does attributes, one per role.

apply statements can be freely mixed with other statements inside the body of the class. In particular, an apply statement that adds fields or methods may appear before or after the class has defined some of its own. It is not required that they appear first.

field

   field $var;
   field @var;
   field %var;

   field $var :ATTR ATTR...;

   field $var = EXPR;

   field $var //= EXPR;
   field $var ||= EXPR;

   field $var { BLOCK }

Since version 0.66.

Declares that the instances of the class or role have a member field of the given name. This member field will be accessible as a lexical variable within any method declarations and ADJUST blocks in the class.

Array and hash members are permitted and behave as expected; you do not need to store references to anonymous arrays or hashes.

Member fields are private to a class or role. They are not visible to users of the class, nor inherited by subclasses nor any class that a role is applied to. In order to provide access to them a class may wish to use "method" to create an accessor, or use the attributes such as ":reader" to get one generated.

The following field attributes are supported:

:reader, :reader(NAME)

Since version 0.27.

Generates a reader method to return the current value of the field. If no name is given, the name of the field is used. A single prefix character _ will be removed if present.

   field $x :reader;

   # equivalent to
   field $x;  method x { return $x }

Since version 0.55 these are permitted on any field type, but prior versions only allowed them on scalar fields. The reader method behaves identically to how a lexical variable would behave in the same context; namely returning a list of values from an array or key/value pairs from a hash when in list context, or the number of items or keys when in scalar context.

   field @items :reader;

   foreach my $item ( $obj->items ) { ... }   # iterates the list of items

   my $count = $obj->items;                   # yields count of items

:writer, :writer(NAME)

Since version 0.27.

Generates a writer method to set a new value of the field from its arguments. If no name is given, the name of the field is used prefixed by set_. A single prefix character _ will be removed if present.

   field $x :writer;

   # equivalent to
   field $x;  method set_x { $x = shift; return $self }

Since version 0.28 a generated writer method will return the object invocant itself, allowing a chaining style.

   $obj->set_x("x")
      ->set_y("y")
      ->set_z("z");

Since version 0.55 these are permitted on any field type, but prior versions only allowed them on scalar fields. On arrays or hashes, the writer method takes a list of values to be assigned into the field, completely replacing any values previously there.

:mutator, :mutator(NAME)

Since version 0.27.

Generates an lvalue mutator method to return or set the value of the field. These are only permitted for scalar fields. If no name is given, the name of the field is used. A single prefix character _ will be removed if present.

   field $x :mutator;

   # equivalent to
   field $x;  method x :lvalue { $x }

Since version 0.28 all of these generated accessor methods will include argument checking similar to that used by subroutine signatures, to ensure the correct number of arguments are passed - usually zero, but exactly one in the case of a :writer method.

:accessor, :accessor(NAME)

Since version 0.53.

Generates a combined reader-writer accessor method to set or return the value of the field. These are only permitted for scalar fields. If no name is given, the name of the field is used. A prefix character _ will be removed if present.

This method takes either zero or one additional arguments. If an argument is passed, the value of the field is set from this argument (even if it is undef). If no argument is passed (i.e. scalar @_ is false) then the field is not modified. In either case, the value of the field is then returned.

   field $x :accessor;

   # equivalent to
   field $x;

   method x {
      $x = shift if @_;
      return $x;
   }

:weak

Since version 0.44.

Generated code which sets the value of this field will weaken it if it contains a reference. This applies to within the constructor if :param is given, and to a :writer accessor method. Note that this only applies to automatically generated code; not normal code written in regular method bodies. If you assign into the field variable you must remember to call Scalar::Util::weaken (or builtin::weaken on Perl 5.36 or above) yourself.

:param, :param(NAME)

Since version 0.41.

Sets this field to be initialised automatically in the generated constructor. This is only permitted on scalar fields. If no name is given, the name of the field is used. A single prefix character _ will be removed if present.

Any field that has :param but does not have a default initialisation expression or block becomes a required argument to the constructor. Attempting to invoke the constructor without a named argument for this will throw an exception. In order to make a parameter optional, make sure to give it a default expression - even if that expression is undef:

   field $x :param;          # this is required
   field $z :param = undef;  # this is optional

Any field that has a :param and an initialisation block will only run the code in the block if required by the constructor. If a named parameter is passed to the constructor for this field, then its code block will not be executed.

Values for fields are assigned by the constructor before any BUILD blocks are invoked.

:inheritable

Experimentally since version 0.807 fields may be optionally inherited when deriving a subclass from another. Not every field is allowed to be inherited. This attribute marks a field as being available for subclasses to inherit.

Field Initialiser Expressions

Since version 0.54 a deferred statement block is also permitted, on any field variable type. This permits code to be executed as part of the instance constructor, rather than running just once when the class is set up. Code in a field initialisation block is roughly equivalent to being placed in a BUILD or ADJUST block.

Since version 0.73 this may also be written as a plain expression introduced by an equals symbol (=). This is equivalent to using a block. Note carefully: the equals symbol is part of the field syntax; it is not simply a runtime assignment operator that happens once at the time the class is declared. Just like the block form describe above, the expression is evaluated during the constructor of every instance.

Since version 0.74 this expression may also be written using a defined-or or logical-or assignment operator (//= or ||=). In these case, the default expression will be evaluated and assigned if the caller did not pass a value to the constructor at all, or if the value passed was undef (for //=) or false (for ||=). For most scalar parameters, where undef is not a valid value, you probably wanted to use //= to assign defaults.

   class Action {
      field $timeout :param //= 20;
      ...
   }

   # The default of 20 will apply here too
   my $act = Action->new( timeout => $opts{timeout} );

Note that $self is specifically not visible during an initialiser expression. This is because the object is not yet fully constructed, so it would be dangerous to allow access to it while in this state. However, the __CLASS__ keyword is available, so initialiser expressions can make use of class-based dispatch to invoke class-level methods to help provide values.

Field initialier expressions were originally experimental, but since version 0.800 no longer emit experimental warnings.

Since version 0.806 fields already declared in a class are visible during the initialisation expression of later fields, and their assigned value can be used here. If the earlier field had a :param declaration, it will have been assigned from the value passed to the constructor. Note however that all ADJUST blocks happen after field initialisation expressions, so any modified values set in such blocks will not be visible at this time.

Control flow that attempts to leave a field initialiser expression or block is not permitted. This includes any return expression, any next/last/redo outside of a loop, with a dynamically-calculated label expression, or with a label that it doesn't appear in. goto statements are also currently forbidden, though known-safe ones may be permitted in future.

Loop control expressions that are known at compiletime to affect a loop that they appear within are permitted.

   field $x { foreach(@list) { next; } }       # this is fine

   field $x { LOOP: while(1) { last LOOP; } }  # this is fine too

has

   has $var;
   has @var;
   has %var;

   has $var = EXPR;

   has $var { BLOCK }

A now-deprecated older version of the "field" keyword.

This generally behaves like field, except that inline expressions are evaluated immediately, once, during class declaration time. These are not stored to be evaluated for each constructor.

Because of the one-shot immediate nature of these initialisation expressions (and a bunch of other reasons), the has keyword is now discouraged for use and will emit compile-time warnings in the deprecated category. Use the field keyword instead.

If you need to evaluate an expression exactly once during the class declaration and assign its now-constant value to every instace, store it in a regular my variable instead:

   my $default_var = EXPR;
   field $var = $default_var;

method

   method NAME {
      ...
   }

   method NAME (SIGNATURE) {
      ...
   }

   method NAME :ATTRS... {
      ...
   }

   method NAME;

Declares a new named method. This behaves similarly to the sub keyword, except that within the body of the method all of the member fields are also accessible. In addition, the method body will have a lexical called $self which contains the invocant object directly; it will already have been shifted from the @_ array.

If the method has no body and is given simply as a name, this declares a required method for a role. Such a method must be provided by any class that implements the role. It will be a compiletime error to combine the role with a class that does not provide this.

The signatures feature is automatically enabled for method declarations. In this case the signature does not have to account for the invocant instance; that is handled directly.

   method m ($one, $two) {
      say "$self invokes method on one=$one two=$two";
   }

   ...
   $obj->m(1, 2);

A list of attributes may be supplied as for sub. The most useful of these is :lvalue, allowing easy creation of read-write accessors for fields (but see also the :reader, :writer and :mutator field attributes).

   class Counter {
      field $count;

      method count :lvalue { $count }
   }

   my $c = Counter->new;
   $c->count++;

Every method automatically gets the :method attribute applied, which suppresses warnings about ambiguous calls resolved to core functions if the name of a method matches a core function.

The following additional attributes are recognised by Object::Pad directly:

:override

Since version 0.29.

Marks that this method expects to override another of the same name from a superclass. It is an error at compiletime if the superclass does not provide such a method.

:common

Since version 0.62.

Marks that this method is a class-common method, instead of a regular instance method. A class-common method may be invoked on class names instead of instances. Within the method body there is a lexical $class available, rather than $self. Because it is not associated with a particular object instance, a class-common method cannot see instance fields.

method (lexical)

   method $var { ... }

   method $var :ATTRS... (SIGNATURE) { ... }

Since version 0.59.

Declares a new lexical method. Lexical methods are not visible via the package namespace, but instead are stored directly in a lexical variable (with the same scoping rules as regular my variables). These can be invoked by subsequent method code in the same block by using $self->$var(...) method call syntax.

   class WithPrivate {
      field $var;

      # Lexical methods can still see instance fields as normal
      method $inc_var { $var++; say "Var was incremented"; }
      method $dec_var { $var--; say "Var was decremented"; }

      method bump {
         $self->$inc_var;
         say "In the middle";
         $self->$dec_var;
      }
   }

   my $obj = WithPrivate->new;

   $obj->bump;

   # Neither $inc_var nor $dec_var are visible here

This effectively provides the ability to define private methods, as they are inaccessible from outside the block that defines the class. In addition, there is no chance of a name collision because lexical variables in different scopes are independent, even if they share the same name. This is particularly useful in roles, to create internal helper methods without letting those methods be visible to callers, or risking their names colliding with other named methods defined on the consuming class.

BUILD

   BUILD {
      ...
   }

   BUILD (SIGNATURE) {
      ...
   }

Since version 0.27.

Declares the builder block for this component class. A builder block may use subroutine signature syntax, as for methods, to assist in unpacking its arguments. A build block is not a subroutine and thus is not permitted to use subroutine attributes (for example :lvalue).

Note that a BUILD block is a named phaser block and not a method. Attempts to create a method named BUILD (i.e. with syntax method BUILD {...}) will fail with a compiletime error, to avoid this confusion.

ADJUST

   ADJUST {
      ...
   }

Since version 0.43.

Declares an adjust block for this component class. This block of code runs within the constructor, after any BUILD blocks and automatic field value assignment. It can make any final adjustments to the instance (such as initialising fields from calculated values).

An adjust block is not a subroutine and thus is not permitted to use subroutine attributes (except see below). Note that an ADJUST block is a named phaser block and not a method; it does not use the sub or method keyword. But, like with method, the member fields are accessible within the code body, as is the special $self lexical.

Currently, an ADJUST block receives a reference to the hash containing the current constructor arguments, as per "ADJUSTPARAMS" (see below). This was added in version 0.66 but will be removed again as it conflicts with the more flexible and generally nicer named-parameter ADJUST :params syntax (see below). Such uses should be considered deprecated. A warning will be printed to indicate this whenever an ADJUST block uses a signature. This warning can be quieted by using ADJUSTPARAMS instead. Additionally, a warning may be printed on code that attempts to access the params hashref via the @_ array.

Since version 0.801 in a future version of this module, ADJUST blocks may be implemented as true blocks and will not permit out-of-block control flow. At present, they are implemented as one full CV per block, but a warning is emitted if out-of-block control flow is attempted.

   ADJUST {
      return;
   }

   Using return to leave an ADJUST block is discouraged and will be removed
   in a later version at FILE line LINE.

Since version 0.805 an experimental feature can be enabled that puts all the ADJUST blocks into a single CV, rather than creating one CV for every block. This is currently being tested for stability, and may become the default behaviour in a future version. For now it must be requested specially:

   use Object::Pad ':experimental(composed_adjust)';

ADJUST :params

   ADJUST :params ( :$var1, :$var2, ... ) {
      ...
   }

   ADJUST :params ( :$var1, :$var2, ..., %varN ) {
      ...
   }

Since version 0.70.

An ADJUST block can marked with a :params attribute, meaning that it consumes additional constructor parameters by assigning them into lexical variables.

Before the block itself, a list of lexical variables are introduced, inside parentheses. The name of each one is preceeded by a colon, and consumes a constructor parameter of the same name. These parameters are considered "consumed" for the purposes of a :strict(params) check.

A named parameter may be provided with default expression, which is evaluated if no matching named argument is provided to the constructor. As with fields, if a named parameter has no defaulting expression it becomes a required argument to the constructor; an exception is thrown by the constructor if it absent.

For example,

   ADJUST :params ( :$x, :$y = "default", :$z ) { ... }

Note here that x and z are required parameters for the constructor of a class containing this block, but y is an optional parameter whose value will be filled in by the expression if not provided. Because these parameters are named and not positional, there is no ordering constraint; required and optional parameters can be freely mixed.

Optional parameters can also use the //= and ||= operators to provide a default expression. In these cases, the default will be applied if the caller did not provide the named argument at all, or if the provided value was not defined (for //=) or not true (for ||=).

   ADJUST :params ( :$name //= "unnamed" ) { ... }

Like with subroutine signature parameters, every declared named parameter is visible to the defaulting expression of all the later ones. This permits values to be calculated based on other ones. For example,

   ADJUST :params ( :$thing = undef, :$things = [ $thing ] ) {
      # Here, @$things is a list of values
   }

This permits the caller to pass a list of values via an array reference in the things parameter, or a single value in thing.

The final element may be a regular hash variable. This requests that all remaining named parameters are made available inside it. The code in the block should delete from this hash any parameters it wishes to consume, as with the earlier case above.

It is unspecified whether named fields or parameters for subclasses yet to be processed are visible to hashes of earlier superclasses. In the current implementation they are, but code should not rely on this fact.

Note also that there must be a space between the :params attribute and the parentheses holding the named parameters. If this space is not present, perl will parse the parentheses as if they are the value to the :params() attribute, and this will fail to parse as intended. As with other attributes and subroutine signatures, this whitespace is significant.

(This notation is borrowed from a plan to add named parameter support to perl's subroutine signature syntax).

ADJUSTPARAMS

Since version 0.51.

   ADJUSTPARAMS ( $params ) {    # on perl 5.26 onwards
      ...
   }

   ADJUST {
      my $params = shift;
      ...
   }

A variant of an ADJUST block that receives a reference to the hash containing the current constructor parameters. This hash will not contain any constructor parameters already consumed by ":param" declarations on any fields, but only the leftovers once those are processed.

The code in the block should delete from this hash any parameters it wishes to consume. Once all the ADJUST blocks have run, any remaining keys in the hash will be considered errors, subject to the ":strict(params)" check.

__CLASS__

   my $classname = __CLASS__;

Since version 0.72.

Only valid within the body (or signature) of a method, an ADJUST block, or the initialising expression of a field. Yields the class name of the instance that the method, block or expression is invoked on.

This is similar to the core perl __PACKAGE__ constant, except that it cares about the dynamic class of the actual instance, not the static class the code belongs to. When invoked by a subclass instance that inherited code from its superclass it yields the name of the class of the instance regardless of which class defined the code.

For example,

   class BaseClass {
      ADJUST { say "Constructing an instance of " . __CLASS__; }
   }

   class DerivedClass :isa(BaseClass) { }

   my $obj = DerivedClass->new;

Will produce the following output

   Constructing an instance of DerivedClass

This is particularly useful in field initialisers for invoking (constant) methods on the invoking class to provide default values for fields. This way a subclass could provide a different value.

   class Timer {
      use constant DEFAULT_DURATION => 60;
      field $duration = __CLASS__->DEFAULT_DURATION;
   }

   class ThreeMinuteTimer :isa(Timer) {
      use constant DEFAULT_DURATION => 3 * 60;
   }

requires

   requires NAME;

Declares that this role requires a method of the given name from any class that implements it. It is an error at compiletime if the implementing class does not provide such a method.

This form of declaring a required method is now vaguely discouraged, in favour of the bodyless method form described above.

CREPT FEATURES

While not strictly part of being an object system, this module has nevertheless gained a number of behaviours by feature creep, as they have been found useful.

Implied Pragmata

The following behaviour is likely to be removed in a later version of this module.

In order to encourage users to write clean, modern code, the body of the class block currently acts as if the following pragmata are in effect:

   use strict;
   use warnings;
   no indirect ':fatal';  # or  no feature 'indirect' on perl 5.32 onwards
   use feature 'signatures';

This behaviour was designed early around the original "line-0" version of the Perl 7 plan, which has subsequently been found to be a bad design and abandoned. That leaves this module in an unfortunate situation, because its behaviour here does not match the plans for core perl; where the recently-added class keyword does none of this, although the method keyword always behaves as if signatures were enabled anyway.

It is eventually planned that this behaviour will be removed from Object::Pad entirely (except for enabling the signatures feature). While that won't in itself break any existing code, it would mean that code which previously ran with the protection of strict and warnings would now not be. A satisfactory solution to this problem has not yet been found, but until then it is suggested that code using this module remembers to explicitly enable this set of pragmata before using the class keyword.

A handy way to do this is to use the use VERSION syntax; v5.36 or later will already perform all of the pragmata listed above.

   use v5.36;

If you import this module with a module version number of 0.800 or higher it will enable a warning if you forget to enable strict and warnings before using the class or roll keywords:

   use Object::Pad 0.800;

   class X { ... }

   class keyword enabled 'use strict' but this will be removed in a later version at FILE line 3.
   class keyword enabled 'use warnings' but this will be removed in a later version at FILE line 3.

Yield True

The following behaviour is likely to be removed in a later version of this module.

A class statement or block will yield a true boolean value. This means that it can be used directly inside a .pm file, avoiding the need to explicitly yield a true value from the end of it.

As with the implied pragmata above, this behaviour has also been found to be a bad design and will likely be removed soon. For now it is suggested not to rely on it and instead either use the new module_true feature already part of the use v5.38 pragma, or on older perls simply remember to put an explicit true value at the end of the file.

SUBCLASSING CLASSIC PERL CLASSES

There are a number of details specific to the case of deriving an Object::Pad class from an existing classic Perl class that is not implemented using Object::Pad.

Storage of Instance Data

Instances will pick either the :repr(HASH) or :repr(magic) storage type.

Object State During Methods Invoked By Superclass Constructor

It is common in classic Perl OO style to invoke methods on $self during the constructor. This is supported here since Object::Pad version 0.19. Note however that any methods invoked by the superclass constructor may not see the object in a fully consistent state. (This fact is not specific to using Object::Pad and would happen in classic Perl OO as well). The field initialisers will have been invoked but the BUILD and ADJUST blocks will not.

For example; in the following

   package ClassicPerlBaseClass {
      sub new {
         my $self = bless {}, shift;
         say "Value seen by superconstructor is ", $self->get_value;
         return $self;
      }
      sub get_value { return "A" }
   }

   class DerivedClass :isa(ClassicPerlBaseClass) {
      field $_value = "B";
      ADJUST {
         $_value = "C";
      }
      method get_value { return $_value }
   }

   my $obj = DerivedClass->new;
   say "Value seen by user is ", $obj->get_value;

Until the ClassicPerlBaseClass::new superconstructor has returned the ADJUST block will not have been invoked. The $_value field will still exist, but its value will be B during the superconstructor. After the superconstructor, the BUILD and ADJUST blocks are invoked before the completed object is returned to the user. The result will therefore be:

   Value seen by superconstructor is B
   Value seen by user is C

STYLE SUGGESTIONS

While in no way required, the following suggestions of code style should be noted in order to establish a set of best practices, and encourage consistency of code which uses this module.

$VERSION declaration

While it would be nice for CPAN and other toolchain modules to parse the embedded version declarations in class statements, the current state at time of writing (June 2020) is that none of them actually do. As such, it will still be necessary to make a once-per-file $VERSION declaration in syntax those modules can parse.

Further note that these modules will also not parse the class declaration, so you will have to duplicate this with a package declaration as well as a class keyword. This does involve repeating the package name, so is slightly undesirable.

It is hoped that eventually upstream toolchain modules will be adapted to accept the class syntax as being sufficient to declare a package and set its version.

See also

File Layout

Begin the file with a use Object::Pad line; ideally including a minimum-required version. This should be followed by the toplevel package and class declarations for the file. As it is at toplevel there is no need to use the block notation; it can be a unit class.

There is no need to use strict or apply other usual pragmata; these will be implied by the class keyword.

   use Object::Pad 0.16;

   package My::Classname 1.23;
   class My::Classname;

   # other use statements

   # field, methods, etc.. can go here

Field Names

Field names should follow similar rules to regular lexical variables in code - lowercase, name components separated by underscores. For tiny examples such as "dumb record" structures this may be sufficient.

   class Tag {
      field $name  :mutator;
      field $value :mutator;
   }

In larger examples with lots of non-trivial method bodies, it can get confusing to remember where the field variables come from (because we no longer have the $self->{ ... } visual clue). In these cases it is suggested to prefix the field names with a leading underscore, to make them more visually distinct.

   class Spudger {
      field $_grapefruit;

      ...

      method mangle {
         $_grapefruit->peel; # The leading underscore reminds us this is a field
      }
   }

WITH OTHER MODULES

Syntax::Keyword::Dynamically

A cross-module integration test asserts that dynamically works correctly on object instance fields:

   use Object::Pad;
   use Syntax::Keyword::Dynamically;

   class Container {
      field $value = 1;

      method example {
         dynamically $value = 2;
         ,..
         # value is restored to 1 on return from this method
      }
   }

Future::AsyncAwait

As of Future::AsyncAwait version 0.38 and Object::Pad version 0.15, both modules now use XS::Parse::Sublike to parse blocks of code. Because of this the two modules can operate together and allow class methods to be written as async subs which await expressions:

   use Future::AsyncAwait;
   use Object::Pad;

   class Example
   {
      async method perform ($block)
      {
         say "$self is performing code";
         await $block->();
         say "code finished";
      }
   }

These three modules combine; there is additionally a cross-module test to ensure that object instance fields can be dynamically set during a suspended async method.

Devel::MAT

When using Devel::MAT to help analyse or debug memory issues with programs that use Object::Pad, you will likely want to additionally install the module Devel::MAT::Tool::Object::Pad. This will provide new commands and extend existing ones to better assist with analysing details related to Object::Pad classes and instances of them.

   pmat> fields 0x55d7c173d4b8
   The field AV ARRAY(3)=NativeClass at 0x55d7c173d4b8
   Ix Field   Value
   0  $sfield SCALAR(UV) at 0x55d7c173d938 = 123
   ...

   pmat> identify 0x55d7c17606d8
   REF() at 0x55d7c17606d8 is:
   └─the %hfield field of ARRAY(3)=NativeClass at 0x55d7c173d4b8, which is:
   ...

DESIGN TODOs

The following points are details about the design of pad field-based object systems in general:

  • Is multiple inheritance actually required, if role composition is implemented including giving roles the ability to use private fields?

  • Consider the visibility of superclass fields to subclasses. Do subclasses even need to be able to see their superclass's fields, or are accessor methods always appropriate?

    Concrete example: The $self->{split_at} access that Tickit::Widget::HSplit makes of its parent class Tickit::Widget::LinearSplit.

IMPLEMENTATION TODOs

These points are more about this particular module's implementation:

  • Consider multiple inheritance of subclassing, if that is still considered useful after adding roles.

  • Work out why no indirect doesn't appear to work properly before perl 5.20.

  • Work out why we don't get a Subroutine new redefined at ... warning if we

      sub new { ... }
  • The local modifier does not work on field variables, because they appear to be regular lexicals to the parser at that point. A workaround is to use Syntax::Keyword::Dynamically instead:

       use Syntax::Keyword::Dynamically;
    
       field $loglevel;
    
       method quietly {
          dynamically $loglevel = LOG_ERROR;
          ...
       }

FEEDBACK

The following resources are useful forms of providing feedback, especially in the form of reports of what you find good or bad about the module, requests for new features, questions on best practice, etc...

SPONSORS

With thanks to the following sponsors, who have helped me be able to spend time working on this module and other perl features.

Additional details may be found at https://github.com/Ovid/Cor/wiki/Sponsors.

AUTHOR

Paul Evans <leonerd@leonerd.org.uk>