NAME

docs/pdds/pdd17_pmc.pod - Parrot Magic Cookies

VERSION

$Revision$

ABSTRACT

This PDD describes the internal structure and behavior of the Parrot Magic Cookie (PMC) data type.

DESCRIPTION

PMCs implement all internal data types more complex than a simple integer, float, or string, and also the data types of high-level languages. Nothing outside the core of Parrot (in fact, nothing outside the data type's vtable routines) should infer anything about a PMC (hence the Magic part).

This does mean, though, that you need to either know what functions are available and what they do, or have some method of finding out. Parrot defines a standard set of functions that each PMC provides. More complex features are constructed out of these fundamental building blocks.

DESCRIPTION

- PMCs contain both state and behavior
- PMCs can inherit from other PMCs
- PMCs can be composed of low-level roles
- High-level objects can subclass low-level PMCs

IMPLEMENTATION

Internal structure

All PMCs have the form:

    struct PMC {
        UnionVal cache;
        Parrot_UInt flags;
        VTABLE *vtable;
        DPOINTER *data;
        struct PMC_EXT *pmc_ext;
    };

where cache is a UnionVal union:

    typedef union UnionVal {
        struct {
            void * _bufstart;
            size_t _buflen;
        } _b;
        struct {
            DPOINTER* _struct_val;
            PMC*      _pmc_val;
        } _ptrs;
        struct _i {
            INTVAL _int_val;
            INTVAL _int_val2;
        } _i;
        FLOATVAL _num_val;
        struct parrot_string_t * _string_val;
    } UnionVal;

u holds data associated with the PMC. This can be in the form of an integer value, a floating-point value, a string value, or a pointer to other data. u may be empty, since the PMC structure also provides a more general data pointer, but is useful for PMCs which hold only a single piece of data (e.g. PerlInts).

flags holds a set of flags associated with the PMC; these are documented in include/parrot/pobj.h, and are generally only used within the Parrot internals.

vtable holds a pointer to the vtable associated with the PMC. This points to a set of functions, with interfaces described in "Vtable Functions" that implement the basic behaviour of the PMC (i.e. how it behaves under addition, subtraction, cloning etc.)

data holds a pointer to the core data associated with the PMC. This may be null.

pmc_ext points to an extended PMC structure. This has the form:

    struct PMC_EXT {
        PMC *_metadata;
        struct _Sync *_synchronize; # [Note: may be deprecated, see STM]
        PMC *_next_for_GC;
    };

_metadata holds internal PMC metadata. The specification for this has not yet been finalized.

_synchronize is for access synchronization between shared PMCs.

_next_for_GC determines the next PMC in the 'used' list during dead object detection in the GC.

PMCs are not required to have a PMC_EXT structure (i.e. pmc_ext can be null).

PMCs are used to implement the basic data types of the high level languages running on top of Parrot. For instance, a Perl 5 SV will map onto one (or more) types of PMC, while particular Python datatypes will map onto different types of PMC.

Defining PMCs

PMCs are declared by the pmclass keyword:

  pmclass <name> [ <modifier> ... ] {
  }

The modifiers specify core features, such as:

need_ext: Adds the PMC_EXT structure when instantiating the PMC.
abstract: The PMC cannot be instantiated. (By convention, abstract classes are given lower-case names.)
no_init: Don't generate class initialization code (don't set up a vtable for the PMC). Used with abstract.
dynpmc: The PMC is a dynamic PMC, so generate special class initialization code suitable for dynamic loading at runtime.
singleton: The PMC is a singleton, created in the constant PMC pool.
no_ro: Don't create a second read-only vtable for the PMC. (Used, for example, by STM's share_ro().)
is_shared: The PMC is shared (across threads).
extends: Inherit from another PMC. Takes one argument, the name of the PMC to inherit from.
does: Compose a role. Takes one argument, the name of the role to compose. [NOTE: this modifier has been taken from the older feature does which is now called provides.]
resolves: Resolve a conflicting vtable function, method, or attribute from a composed role by using the same named vtable function, method, or attribute from the current pmclass or prole.
provides: The PMC satisfies a particular low-level interface, which gives some assurances on how and where a PMC can be used.; Roles composed with does may also define provides for one or more interfaces. (They generally define at least a provides corresponding to their own name.)
hll: Declare the PMC in an HLL namespace other than the default HLL 'parrot'. Takes one argument, the name of the HLL.
maps: Map the current PMC to a core PMC type for code declared in a particular HLL. May only be used together with the hll modifier.

Defining attributes

The attributes of a PMC (both public and private) live in a custom struct for the PMC, stored in the data member of the PMC struct. The standard way of declaring attributes is with ATTR within the body of a pmclass or prole declaration.

  ATTR <type> <name> [ :<modifier> ... ];

This declaration is used to generate the data struct for the PMC (named "Parrot_<pmcname>_Data"). The data struct incorporates any attributes declared in a composed role or inherited class. Composed attributes are inserted at the end of the generated struct. Inherited attributes are inserted at the top of the generated struct, so the struct members shared by the parent and child are in the same position, and code compiled for direct struct access to the parent can also perform the same direct struct access on the child.

The declaration is also used to generate accessor macros, GET_ATTR_attrname and SET_ATTR_attrname, to set up helper information for the inspect vtable function, and to provide attribute access through the default get_attr and set_attr vtable functions.

It's also possible to define and store the PMC's data struct manually, but the standard declaration syntax must be used in roles, in PMCs that compose roles, and in PMCs that will be subclassed by high-level classes (this means all core PMCs, and most non-core PMCs).

Low-level PMCs that use multiple inheritance (pmclass declarations with more than one extends entry) have to be careful with the contents of their structs. There's no problem if the two parents have identical data struct members. But, if the two parents have different data struct members, any vtable functions inherited from the parents will not be able to directly access the parents' struct members or use the accessor macros. The solution is to either override those vtable functions that directly access data struct members in the child PMC, define the parents as roles instead of classes, or declare the multiply inheriting child in PIR or an HLL.

Defining vtable functions

Vtable functions are defined as C functions within the body of the pmclass declaration, and are marked with VTABLE.

  VTABLE STRING *get_string() {...}
  VTABLE void set_string_native(STRING *value) {...}

Within the body of vtable functions, several shortcuts are provided:

INTERP: The current interpreter object.
SELF: The current invocant by dynamic type.
STATICSELF: The current invocant by static type.
SUPER: Calls the current method in the nearest superclass, using the dynamic type of SELF.
STATICSUPER: Calls the current method in the nearest superclass, using the static type of SELF.

Methods

Methods are declared in the body of the pmclass or prole declaration with METHOD.

  METHOD inspect(STRING *what :optional, int got_what :opt_flag) {...}

PMCs and Namespaces

Like high-level classes, low-level PMCs are tied to a corresponding namespace. By default this is a namespace with the same name as the PMC in the 'parrot' HLL, but the hll modifier in the pmclass declaration selects a different HLL.

Accessing a core PMC type from within an HLL other than 'parrot' requires the same steps as accessing a class from another HLL, first retrieving the namespace object, and then instantiating from that namespace.

  $P0 = get_root_namespace ['parrot'; 'Integer']
  $P1 = new $P0

HLLs can choose to provide direct access to Parrot's core PMC types by aliasing them within the HLL namespace.

Inheritance

PMCs can inherit behavior and state from other PMCs. Inheritance is performed in the pmclass declaration using the extends keyword.

  pmclass Foo extends Bar {
  }

Composition

Composition is another form of code reuse in PMCs. Unlike inheritance, composed roles aren't complete stand-alone PMCs, they are just bundles of behavior and state that can be used within the composing PMC. As such, roles are never instantiated directly, and are never translated to C directly. They have no core structs, though they define attributes to be added to the PMC they are composed into.

When a PMC that uses a role is translated to C, the role provides vtable functions, methods, and attributes that will be added to the generated C code for that PMC. Composed vtable functions, methods, and attributes are not permitted to have the same name as corresponding features defined in the composing PMC. This is called a conflict, and must be explicitly resolved in the composing PMC. When composed features have the same name as inherited vtable functions, methods, or attributes, the composed feature overrides the inherited feature, just as it would if defined in the composing PMC.

Roles are defined using the prole keyword.

  prole <name> <modifiers> {
  }

Roles can compose other roles, but they can't inherit.

Role attributes are defined using the same format as PMC attributes.

Core roles live in src/role and have a file extension of .pr.

Roles are composed into a PMC with the does modifier.

PMCs and high-level objects

High-level objects, as specified in PDD15, need to be able to inherit from PMCs. Subclassing a low-level PMC from a high-level class makes an entry in the high-level class's list of parents.

For PDD15 objects, there is a corresponding instance of the Class PMC. For a low-level PMC, however, the class definition is written in C and compiled away. There needs to be something placed in the parents list for a PDD15 class, that can provide access to the low-level PMC's vtable and methods, and define the storage that the low-level PMC will need within the high-level object. That something is the PMCProxy PMC. Like a PDD15 class, it is stored as the class element in the namespace associated with a PMC, provides introspection facilities and can sit in an inheritance hierarchy.

The PMCProxy PMCs are only created when needed for subclassing a low-level PMC, to avoid a large load of unused PMCProxy objects. When created, they are cached in the class slot of the namespace corresponding to the low-level PMC, so they are only created once.

Therefore, subclassing a PMC looks, at a PIR level, like subclassing a high level class.

  $P0 = get_class 'Hash'
  $P1 = newclass 'MyClass'
  addparent $P1, $P0     # The new class inherits from the Hash PMC

Or, more briefly:

  $P1 = subclass 'Hash', 'MyClass'

PMCs store state in a very different way to PDD15 objects. When a method inherited from a PMC is called on a PDD15 object, that method needs to access the attributes of the inherited low-level PMC. Further, if multiple PMCs are inherited from, they may each have attributes with the same name, that need to be correctly "visible" to the PDD 15 object according to the laws of inheritance. Users of Parrot at a PIR level should not have to care about such issues.

To enable attributes from the low-level PMC to act as full inherited attributes in the child class, the PMCProxy class will create a set of PDD 15 attributes that correspond in type and name to the attributes declared with ATTR in the declaration body of the low-level PMC, as if each had been added with add_attribute. It will also override the GET_ATTR_attrname and SET_ATTR_attrname functions to point to the PDD 15 attributes (with automatic boxing and unboxing for the PMC values) rather than to members of a C struct.

The PMCProxy will also scan the low-level PMC for methods declared with METHOD and insert them in the proxy class as if each had been declared with add_method (possibly with a shim to standardize calling conventions, but hopefully the calling conventions are similar enough between C-defined and PIR-defined methods not to need a shim). The PMCProxy will maintain a link to the low-level PMC's vtable, and use it for any vtable calls that aren't overridden by the proxy class itself.

As a result, a low-level PMC used as a parent of a PDD 15 class will never be instantiated directly. It will only be used as a source for attribute names and types, methods, and a vtable.

When a method is called on an object whose class has low-level PMC parents, the call is made exactly as it would be for PDD 15 parents. The invocant is always the PDD 15 object. Any method or vtable calls made within the low-level PMC are dispatched on the PDD 15 object invocant. This allows the PDD 15 object to intelligently handle method and vtable overrides within multiple parents and itself.

If a low-level PMC expects to be overridden by high-level classes (which means all the core low-level PMC types), it must respect the standard interface.

REFERENCE

Vtable Functions

Vtables decouple the interface and implementation of various object functions. The actual vtable structure contains pointers to functions that implement the methods for that particular PMC. All pointers must point to valid functions with appropriate prototypes.

In C code, the first parameter to any vtable routine is the current interpreter. The second parameter is the PMC itself.

The following list details each of the vtable functions, their prototypes, and their behavior.

Core Vtable Functions

init: Called when a PMC is first instantiated. It takes an unused PMC parameter and turns it into a PMC of the appropriate class.
init_pmc: Alternative entry point called when a PMC is first instantiated. Accepts a PMC parameter used to initialize the given object. Interpretation of the PMC initializer is left open, each PMC is free to choose its own implemention. A null value passed as the initializer parameter is allowed.; NOTE: It is strongly suggested that init_pmc(PMCNULL) be equivalent to init(), though there will of necessity be exceptions.
instantiate: Creates a new PMC object of the type of the class and calls init_pmc(), passing in the initialization argument, init, which is usually a hash. This opcode is most often used with high-level classes, but low-level PMCs may instantiate an object of their own type.
new_from_string [deprecated: See RT# 47011]: Construct a PMC initializing it from a string representation and integer flags. [NOTE: this has been replaced by instantiate. Any initialization arguments can be passed in the init hash.]
inspect: Return a hash of all characteristics of the self PMC available for introspection.; Return a PMC value for one characteristic of the self PMC, selected by string name. Returns PMCNULL if the characteristic doesn't exist.
morph: Turn the PMC into a PMC of type type. If the morphing can't be done in any reasonable way -- for instance if an integer is asked to turn into an Array -- then the PMC is first destroyed, then recreated as an empty PMC of the new type.; This method is primarily used when the interpreter has need of coercing a PMC to a particular type, and isn't meant as a general purpose casting tool. Compilers should only emit valid morphing operations.
mark: Called when the DOD is tracing live PMCs. If this method is called then the code must mark all strings and PMCs that it contains as live, otherwise they may be collected.; This method is only called if the DOD has detected that this PMC is both alive and has a custom mark routine as indicated by the custom mark PMC flag. (Most normal PMCs don't need a custom mark routine.); If a PMC has this flag set, then it is responsible for marking all buffers and PMCs under its control as alive. If it does not, those PMCs or buffers may be collected later. This method does not have to call the mark method on any PMCs it marks--the DOD system takes care of that. (So no need to recurse into aggregate PMCs or anything of the sort).; This method may allocate no memory from Parrot, nor may it alter Parrot's internal structures. It should have no side-effects from the C level either. This routine may not throw an exception.
destroy: Called when the PMC is destroyed. This method is called by the DOD when it determines that a PMC is dead and that the PMC has marked itself as having a destroy method (an active finalizer).; When this method finishes, the PMC will be marked as dead. As such you should make sure that you don't leave any references to it in any Parrot structure by the end of the method.; This method may not throw an exception. It will be ignored if it does.
clone: Return a clone of a PMC.
defined: Return a true value if the PMC is defined, false otherwise.
get_class: Return the class object for this PMC. For high-level objects, this is a Class object (or other high-level class object). For low-level PMCs, this returns a PMCProxy object.
pmc_namespace [deprecated: See RT# 48144]: Return the namespace object for this PMC. [NOTE: replaced by get_namespace.]
get_namespace: Return the namespace object for this PMC.
freeze: Freeze the PMC to an archived string format (a bytecode string constant that can be saved in a packfile).
thaw: Thaw a PMC from an archived string format (a bytecode string constant that can be saved in a packfile).
thawfinish: Called after the PMC has been thawed to perform any finalization steps.
visit: Used by freeze and thaw to visit the contents of the PMC.
share: Mark a PMC as shared and read-only.; [NOTE: these seem to be used interchangably, should be distinguished or merged.]

Accessors

getprop: Return the value from the property hash of self keyed by key. The key should not be null.
setprop: Set the value in the property hash of self that is keyed by key to the value of value. The key should not be null.
delprop: Delete the value from the property hash of self keyed by key. The key should not be null.
getprops: Return the entire property hash for self.
type [deprecated: See RT #48567]: Return the type of the PMC. Type is a unique number associated with the PMC when the PMC's class is loaded. Negative numbers are considered interpreter-specific, non-public types.
subtype [deprecated: See RT #48569]: Return the subtype of a PMC. (Note that this may be unimplemented, and may go away). This is intended to return information about the PMC--what type of number or string it is, whether it's a scalar, hash, array, or list, and suchlike things.; [This can be adequately handled by does and inspect.]
name: Return the name of the class for the PMC.
get_integer: Return the native integer value of the PMC.
get_number: Return the native floating-point value of the PMC.
get_bignum: Return the extended precision numeric value of the PMC as a new bignum PMC.
get_string: Return the native string value of the PMC. This may be in any encoding, chosen by the PMC.
get_repr: Return a string representation of the PMC. [NOTE: redundant with get_string, candidate for deprecation.]
get_bool: Return the true/false value of the PMC (the constant TRUE or the constant FALSE). The definition of truth for a given PMC will depend on the type of the PMC. For a scalar, it may be as simple as returning false when the PMC has a value 0 or "", and returning true when the PMC has any other value.
get_pmc: Return the PMC value for this PMC. This is useful in circumstances where the thing being accessed may return something other than its own value. For example, an array might return a reference to itself. Any PMC may return a value different from the PMC that get_pmc is being called on.
set_integer_native: Set the integer value of this PMC from a native integer value (integer register/constant).
set_integer_same: Set the value of this PMC from the integer value of another PMC. The value PMC is guaranteed to be of the same type as the self PMC, so optimizations may be made.
set_number_native: Set the value of this PMC from a native floating-point value (float register/constant).
set_number_same: Set the value of this PMC from the floating-point value another PMC. The value PMC is guaranteed to be of the same type as the self PMC, so optimizations may be made.
get_pointer: Returns a pointer value for the PMC. Useful for PMCs that hold pointers to arbitrary data. The details of the data (type, location etc.) depend on the PMC.
set_bignum_int: Morph the PMC to a BIGNUM PMC, and set the extended-precision value from a native integer.
set_string_native: Set the value of this PMC from a native string value (string register/constant).
assign_string_native: Set the value of this PMC to a copied native string value (string register/constant).
set_string_same: Set the value of this PMC from the string value of another PMC. The value PMC is guaranteed to be of the same type as the self PMC, so optimizations may be made.
set_bool: Set the boolean state of the PMC to TRUE if the native integer value passed in is TRUE, or FALSE if the value is FALSE. The definition of truth is left open to the particular PMC. For a scalar, it may be as simple as setting false when a 0 value is passed in, and seting true when any other value is passed in.
assign_pmc: Set the value of the PMC in self to the value of the PMC in value by copying the value.
set_pmc: Make the PMC in self refer to the PMC passed as value.
set_pointer: Set the pointer value of the PMC Useful for PMCs that hold pointers to arbitrary data. The details of the data (type, location etc.) depend on the PMC.

Aggregate Vtable Functions

Many of the following functions have a *_keyed form, a *_keyed_int form, and a *_keyed_str form. The keyed forms take a PMC*, INTVAL, or STRING* key as a parameter. The PMC* parameter is null (PMCNULL) if there is no key for that PMC; this means that that argument is unkeyed.

In some cases, the caller must provide a non-null key. Those cases are explicitly stated below. In the other cases, you may have to implement the keyed vtable functions and check for a null self key even if you are implementing a non-aggregate type. If the self key is non-null and the PMC class is a non-aggregate type, the _keyed_* methods should throw an exception.

If you do not implement the *_keyed_int and *_keyed_str functions, the default will convert the INTVAL or STRING* into a key PMC* and call the corresponding *_keyed functions.

elements: Return the number of elements in the PMC.
get_integer_keyed: Return the integer value for the element indexed by a PMC, integer, or string key. The key is guaranteed not to be null for this function.
get_number_keyed: Return the native floating-point value for the element indexed by a PMC, integer, or string key. The key is guaranteed not to be null for this function.
get_string_keyed: Return the string value for the element indexed by a PMC, integer, or string key. The key is guaranteed not to be null for this function.
get_bool_keyed [deprecated: See RT #48571]: Return the boolean value for the element indexed by a PMC key.
get_bool_keyed_int [deprecated: See RT #48573]: Return the boolean value for the element indexed by an integer key.
get_bool_keyed_str [deprecated: See RT #48575]: Return the boolean value for the element indexed by a string key.
get_pmc_keyed: Return the PMC value for the element indexed by a PMC, integer, or string key. The key is guaranteed not to be null for this function.
get_pointer_keyed: Return the pointer value for the element indexed by a PMC, integer, or string key. The details of the data (type, location etc.) depend on the PMC.
set_integer_keyed: Set the integer value of the element indexed by a PMC, integer, or string key. The key is guaranteed not to be null for this function.
set_number_keyed: Set the floating-point value of the element indexed by a PMC, integer, or string key. The key is guaranteed not to be null for this function.
set_string_keyed: Set the string value of the element indexed by a PMC, integer, or string key. The key is guaranteed not to be null for this function.
set_pmc_keyed: Set the PMC value of the element indexed by a PMC, integer, or string key by copying the value of another PMC.
set_pointer_keyed: Set the pointer value of the element indexed by a PMC, integer, or string key.
type_keyed [deprecated: see individual tickets below]: Return the type number of the PMC indexed by a PMC, integer, or string key. The key parameter is guaranteed not to be null for this method.
pop_integer: Return the integer value of the last item on the list, removing that item.
pop_float: Return the floating-point value of the last item on the list, removing that item.
pop_string: Return the string value of the last item on the list, removing that item.
pop_pmc: Return the PMC value of the last item on the list, removing that item.
push_integer: Add the passed in integer value to the end of the list.
push_float: Add the passed in floating-point number to the end of the list.
push_string: Add the passed in string to the end of the list.
push_pmc: Add the passed in PMC to the end of the list.
shift_integer: Return the integer value of the first item on the list, removing that item.
shift_float: Return the floating-point value of the first item on the list, removing that item.
shift_string: Return the string value of the first item on the list, removing that item.
shift_pmc: Return the PMC value of the first item on the list, removing that item.
unshift_integer: Add the passed in integer value to the beginning of the list.
unshift_float: Add the passed in floating-point number to the beginning of the list.
unshift_string: Add the passed in string to the beginning of the list.
unshift_pmc: Add the passed in PMC to the beginning of the list.
splice: Replace some number of PMCs (specified by the integer count) at offset offset from the beginning of self with the PMCs in the aggregate value.
exists_keyed: Check if the element indexed by a PMC, integer, or string key exists.
defined_keyed: Check if the element indexed by a PMC, integer, or string key is defined.
delete_keyed: Delete the element indexed by a PMC, integer, or string key.
nextkey_keyed: Advance to the next position while iterating through an aggregate. [NOTE: this feature needs review together with the Iterator PMC.]

Math Vtable Functions

add: Add the value of self to the value of a PMC, native integer, or native floating-point number and store the result in a PMC dest. Note that dest may be the same PMC as self; in that case optimizations may be made. The i_ variants perform an inplace operation, modifying the value of self.
subtract: Subtract the value of a PMC, native integer, or native floating-point number from a PMC and store the result in dest. If dest is null create a result PMC of an appropriate type. Note that dest may be the same PMC as self; in that case optimizations may be made. The i_ variants perform an inplace operation, modifying the value of self.
increment: Increment the value of a PMC by 1.
decrement: Decrement the value of a PMC by 1.
multiply: Multiply a PMC, native integer, or floating-point value by the value of the PMC self and store the result in the dest PMC. Note that dest may be the same PMC as self; in that case optimizations may be made. The i_ variants perform an inplace operation, modifying the value of self.
divide: Divide the value of the self PMC by a PMC, native integer, or native floating-point number and store the result in dest. Note that dest may be the same PMC as self; in that case optimizations may be made. The i_ variants perform an inplace operation, modifying the value of self.
floor_divide: Divide the PMC's value number by value and return the result in dest. The result is the floor() of the division i.e. the next whole integer towards -inf. If the denominator is zero, a 'Divide by zero' exception is thrown. The i_ variants perform an inplace operation, modifying the value of self.
modulus: Divide the value of the self PMC by the value of a PMC, native integer, or native floating-point number and store the remainder in dest. Note that dest may be the same PMC as self; in that case optimizations may be made. The i_ variants perform an inplace operation, modifying the value of self.
cmodulus: Divide the value of the self PMC by the value of a PMC, native integer, or native floating-point number and store the remainder in dest. Note that dest may be the same PMC as self; in that case optimizations may be made. The i_ variants perform an inplace operation, modifying the value of self.; Note that modulus uses Knuth's "corrected mod" algorithm, as implemented in src/utils.c, while cmodulus uses the C-style fmod function.
pow: Return the value of self raised to the power of value. The i_ variants perform an inplace operation, modifying the value of self.
absolute: Return the absolute value of self. The i_ variant performs an inplace operation, modifying the value of self.
neg: Negate the sign of self and store the result in dest. Note that self and dest may refer to the same PMC, in which case optimizations may be made. The i_ variant performs an inplace operation, modifying the value of self.

Logical Vtable Functions

bitwise_or: Calculate the bitwise-OR of the value of the self PMC and the value of a PMC or native integer and store the result in dest. Note that dest may be the same PMC as self; in that case optimizations may be made. [Question: what happens when the self and value PMCs aren't integers?]; The i_ variants perform an inplace operation and store the result in self.
bitwise_and: Return the result of a bitwise AND on the passed in value and the self PMC. The i_ variants perform an inplace operation and store the result in self.
bitwise_xor: Return the result of a bitwise XOR on the passed in value and the self PMC. The i_ variants perform an inplace operation and store the result in self.
bitwise_ors: Return the result of a bitwise OR over an entire string on the passed in value and the self PMC. The i_ variants perform an inplace operation and store the result in self.
bitwise_ands: Return the result of a bitwise AND over an entire string on the passed in value and the self PMC. The i_ variants perform an inplace operation and store the result in self.
bitwise_xors: Return the result of a bitwise XOR over an entire string on the passed in value and the self PMC. The i_ variants perform an inplace operation and store the result in self.
bitwise_not: Returns the bitwise negation of the self PMC. The i_ variant performs an inplace operation, storing the result in self.
bitwise_nots: Returns the bitwise negation of the string self PMC. The i_ variant performs an inplace operation, storing the result in self.
bitwise_shl: Return the value of the self PMC bitwise shifted left by the amount specified in value, shifting in zeroes on the right (arithmetic/logical bitwise shift). A negative value shifts right. The i_ variants perform an inplace operation, storing the result in self.; The result may be promoted to a BigInt.
bitwise_shr: Return the value of the self PMC bitwise shifted right by the amount specified in value, shifting in copies of the sign bit on the left (arithmetic bitwise shift). A negative value shifts left. The i_ variants perform an inplace operation, storing the result in self.; The result may be promoted to a BigInt (when value is negative).
bitwise_lsr: Return the value of the self PMC bitwise shifted right by the amount specified in value, shifting in zeroes on the left (logical bitwise shift). A negative value shifts left. The i_ variants perform an inplace operation, storing the result in self.
is_equal: Return an integer value (1/0) indicating if the self PMC is equal to the value PMC. The _num version tests for numeric equality, while the _string version tests for string equality.
is_same: Return an integer value (1/0) indicating if self is the same as value (with "sameness" determined by each PMC).
cmp: Returns the integer result of comparing the values of self and value (0 for equal, 1 if self is greater, -1 if value is greater). The _num version performs a numeric comparison, while the _string version performs a string comparison.
logical_or: Performs a logical OR, returning self if it is true, and value otherwise.
logical_and: Performs a logical AND, returning a true value if both self and value are true, and a false value otherwise. (Currently implemented as: If self is false, return it, since the entire expression is false. If self is true, return value, since the truth or falsehood of value will determine the truth or falsehood of the whole expression.)
logical_xor: Performs a logical XOR, returning self if it is true and value is false, returning value if it is true and self is false, and returning a false value (PMC of the same type as self, with the boolean value 0) if both are true or neither are true.
logical_not: Returns the logical negation of self. The i_ variant performs an inplace negation, modifying the value of self.

String Vtable Functions

concatenate: Concatenate self with a PMC or string value and return the result. The i_ variant performs an inplace concatenation, modifying the value of self.
repeat: Return the result of repeating the value in self the number of times indicated in value. The i_ variants perform an inplace operation, modifying the value of self.
substr: Extracts the string starting at offset with size length and return it as a PMC in dest or as a string return value.

Code Vtable Functions

invoke: Invoke the code object self.

Class/Object Vtable Functions

can: Return a true value if the PMC has a method named method, return 0 otherwise.
isa: Return a true value if the PMC inherits from the class named classname, return 0 otherwise.
does: Return a true value if the PMC does (composes) or performs (satisfies the interface of) the role named role, return 0 otherwise.
get_attr: Retrieve an attribute value from the PMC (instance object).
set_attr: Store an attribute value in the PMC (instance object).
add_parent: Add a parent to the PMC (class object), establishing an inheritance relation.
remove_parent: Remove a parent from a PMC (class object).; Not all object metamodels will support removing parents.
add_role: Add a role to the PMC (class object), establishing a composition relation.
remove_role: Remove a role from the PMC (class object).; Not all object metamodels will support removing roles.
add_attribute: Add an attribute to the PMC (class object).
remove_attribute: Remove an attribute from the PMC (class object).; Not all object metamodels will support removing attributes.
add_method: Add a method to the PMC (class object).
remove_method: Remove a method from the PMC (class object).; Not all object metamodels will support removing methods.
add_vtable_override: Add a vtable override to the PMC (class object).; Remove a vtable override from the PMC (class object).
find_method: Return a subroutine PMC for the passed method name. This subroutine PMC may be cached, so the method must return an equivalent sub PMC each time, or be capable of dealing with the returned sub PMCs being reused. [Why should it be cached? Can you turn off caching? What if you want to override find_method to generate methods on the fly?]

Core PMCs

Parrot has a number of core PMC types that all programs can guarantee will be available to them. (With the possible exception of Parrot programs executing on an embedded device or other restricted environment)

Scalar types

Undef: This is the generic no-value type. It has a numeric value of zero, a string value of empty string, and a boolean value of false. It will, on assignment, turn itself into a PMC of the source type, or if assigned a basic type will turn itself into one of the wrapper PMC types (detailed below) for the basic types.
Integer: The PMC wrapper for Parrot's low-level integer type. Always an integer, with other types auto-converted to an integer when stored into this PMC. The range and behaviour of the Integer PMC is identical to the platform low-level integer.; The boolean value for an Integer is false if zero, otherwise true.; Floating point, string, and bignum values assigned to an Integer PMC round to the nearest integer. Floats, or strings which resolve to numbers, cap at the platform maximum or minimum integer value.; Integer PMCs take on a value of 1 if a boolean true is assigned, and a value of 0 if a boolean false is assigned.; If an out-of-range value is assigned to an Integer PMC, the PMC will throw an exception if exact math is enabled.
Float: The PMC wrapper for Parrot's low-level floating-point type. Always a float, with other types autoconverted to a float when stored into this PMC.; The boolean value for a Float is false if exactly zero, otherwise true.; When converted to an integer, floats round to the closest integer, capping at the platform maximum or minimum integer value.; When converting to a string, floats use the platform default snprintf format.
String: The PMC wrapper for Parrot's low-level string type. Always a simple string, with other types autoconverted to a string when stored into this PMC.; The boolean value for a String is false if empty or the string '0' (a one character string holding a zero) otherwise true. This PMC autoconverts to an integer or float when its integer or float value is fetched.
Boolean: A true/false value. Returns 0 for false, 1 for true when fetched as an integer or float, empty string for false and '1' for true when fetched as a string.
BigInt: An arbitrary precision integer.
BigNum: The PMC wrapper for Parrot's low-level BigNum type. {{ NOTE: this type doesn't seem to exist. }}
Complex: A complex number, consisting of a real part and an imaginary part. {{ NOTE: is this a complete and useful implementation of complex numbers? }}
Class: The PMC for Parrot's class.
Object: The PMC for Parrot's base object type.
Ref: The PMC that represents a reference to another PMC. Delegates all functions to the referred-to PMC.
AggregateElementRef: This PMC represents a reference to an element contained in an aggregate PMC type, such as an array or hash. It is initialized with the key being referenced and the aggregate PMC containing that key.; Note that assigning to the reference PMC will be equivalent to a keyed set on the referenced aggregate PMC - that is, it modifies the element rather than doing a v-table call on the element itself. It is important to be aware of this when assigning a PMC through this reference; it is not the same behaviour as the Ref PMC.
WeakRegisterRef: This PMC represents a weak reference to a register. Should the reference live beyond the context containing the register that it references, any attempt to use the reference will throw an exception.; A weak register reference can only be created by the register_ref opcode. Any assignment to the register will behave like a set instruction. That is, when assigning a PMC using a WeakRegisterRef PMC, the register will be updated to reference that PMC rather than calling the assign v-table call on the PMC in that register. This is not the same behaviour as the Ref PMC.
Random: A singleton PMC that generates a random number. {{ NOTE: Why do we have this? }}
Exception: The base class for all exceptions. Currently based on ResizablePMCArray, but that's likely to change.

Array types

Note that for the following types you can set the size of the array by using the VTABLE_set_integer_native() method. Assigning an integer to the array as a whole sets the array to that size.

Size-changing operations (such as push, pop, shift, unshift, and splice) on statically-sized arrays will throw an exception.

ResizablePMCArray returns Undef for unset elements (so does the new object model, because it uses ResizablePMCArray for storage), but Hash returns PMCNULL. Standardize all core aggregate PMC types on the singleton PMCNULL.

Array: The base class for all array types (a statically sized array for any arbitrary type). New array types can be derived from the base Array. In user code it is recommended to use one of the specific array types below, rather than the base type.
FixedBooleanArray: A statically sized array which holds only Boolean values.
ResizableBooleanArray: A dynamically sized array which holds only Boolean values.
FixedIntegerArray: A statically sized array which holds only Integer values.
ResizableIntegerArray: A dynamically sized array which holds only Integer values.
FixedFloatArray: A statically sized array which holds only Float values.
ResizableFloatArray: A dynamically sized array which holds only Float values.
FixedPMCArray: A statically sized array which holds only PMC values.
ResizablePMCArray: A dynamically sized array which holds only PMC values.
FixedStringArray: A statically sized array which holds only String values.
ResizableStringArray: A dynamically sized array which holds only String values.

Hash types

Pair: A Pair PMC represents one key => value mapping. Thus it is like a one element hash or a two element array.
Hash: A container with key-value semantics. The values are PMCs.
Env: Env is a singleton PMC class, that is there is only one Env PMC in any interpreter. This PMC gives access to the process' environment variables--reading from it returns the value of the named process environment variable, while writing to it sets the value of a process environment variable. For example, to retrieve the current value of TERM (the terminal type on most Unix systems):; Note that an embedding system may override this behavior.
NameSpace: Stores one level of a namespace. Every slot in a namespace contains either another namespace (the next level down), or a variable or subroutine/method.
OrderedHash: A hash that also preserves the order of elements, providing the interface of both a hash and an array.
AddrRegistry: Simulates reference counting for dead-object detection and garbage collection.

Subroutine types

Sub: A fundamental subroutine object, and base class for other subroutine PMC types.
Closure: A closure: subroutine object plus captured lexical scope.
Continuation: A continuation: a subroutine object that captures the interpreter's context at the point where the continuation was constructed.
Coroutine: A coroutine: a continuation object that can stop part way through execution and restart at the point where it left off the next time it's called.
Eval: An extension of Sub that provides dynamic code evaluation and execution.
Exception_Handler: A code object for handling exceptions.
MultiSub: A container for multiply dispatched subroutines.
NCI: A native call interface wrapper around a C function.
Bound_NCI: An internal NCI method call bound to a particular call instance.
Compiler: A subroutine implementing a language compiler. (Derived from NCI.)

ATTACHMENTS

None.

FOOTNOTES

None.

REFERENCES

  docs/pmc2c.pod

parrotcode: Parrot Magic Cookies
Contents \| Documentation