The main goal of Logtalk objects is the encapsulation and reuse of predicates. Instead of a single database containing all your code, Logtalk objects provide separated namespaces or databases allowing the partitioning of code in more manageable parts. Logtalk does not aim to bring some sort of new dynamic state change concept to Logic Programming or Prolog.
In Logtalk, the only pre-defined objects are the built-in objects user
and logtalk
, which are described at the end of this section.
There are only three kinds of encapsulation entities in Logtalk: objects, protocols, and categories. Logtalk uses the term object in a broad sense. The terms prototype, parent, class, subclass, superclass, metaclass, and instance always designate an object. Different names are used to emphasize the role played by an object in a particular context. I.e. we use a term other than object when we want to make the relationship with other objects explicit. For example, an object with an instantiation relation with other object plays the role of an instance, while the instantiated object plays the role of a class; an object with a specialization relation with other object plays the role of a subclass, while the specialized object plays the role of a superclass; an object with an extension relation with other object plays the role of a prototype, the same for the extended object. A stand-alone object, i.e. an object with no relations with other objects, is always interpreted as a prototype. In Logtalk, entity relations essentially define patterns of code reuse. An entity is compiled accordingly to the roles it plays.
Logtalk allows you to work from standalone objects to any kind of hierarchy, either class-based or prototype-based. You may use single or multiple inheritance, use or forgo metaclasses, implement reflective designs, use parametric objects, and take advantage of protocols and categories (think components).
Prototypes are either self-defined objects or objects defined as extensions to other prototypes with whom they share common properties. Prototypes are ideal for representing one-of-a-kind objects. Prototypes usually represent concrete objects in the application domain. When linking prototypes using extension relations, Logtalk uses the term prototype hierarchies although most authors prefer to use the term hierarchy only with class generalization/specialization relations. In the context of logic programming, prototypes are often the ideal replacement for modules.
Classes are used to represent abstractions of common properties of sets of objects. Classes provide an ideal structuring solution when you want to express hierarchies of abstractions or work with many similar objects. Classes are used indirectly through instantiation. Contrary to most object-oriented programming languages, instances can be created both dynamically at runtime or defined in a source file like other objects.
We can define a new object in the same way we write Prolog code: by using a text editor. Logtalk source files may contain one or more objects, categories, or protocols. If you prefer to define each entity in its own source file, it is recommended that the file be named after the object. By default, all Logtalk source files use the extension .lgt
but this is optional and can be set in the adapter files. Intermediate Prolog source files (generated by the Logtalk compiler) have, by default, a _lgt
suffix and a .pl
extension. Again, this can be set to match the needs of a particular Prolog compiler in the corresponding adapter file. For instance, we may define an object named vehicle
and save it in a vehicle.lgt
source file which will be compiled to a vehicle_lgt.pl
Prolog file.
Object names can be atoms or compound terms (when defining parametric objects, see below). Objects, categories, and protocols share the same name space: we cannot have an object with the same name as a protocol or a category.
Object code (directives and predicates) is textually encapsulated by using two Logtalk directives: object/1-5
and end_object/0
. The most simple object will be one that is self-contained, not depending on any other Logtalk entity:
:- object(Object). ... :- end_object.
If an object implements one or more protocols then the opening directive will be:
:- object(Object, implements(Protocol)). ... :- end_object.
An object can import one or more categories:
:- object(Object, imports(Category)). ... :- end_object.
If an object both implements protocols and imports categories then we will write:
:- object(Object, implements(Protocol), imports(Category)). ... :- end_object.
In object-oriented programming objects are usually organized in hierarchies that enable interface and code sharing by inheritance. In Logtalk, we can construct prototype-based hierarchies by writing:
:- object(Prototype, extends(Parent)). ... :- end_object.
We can also have class-based hierarchies by defining instantiation and specialization relations between objects. To define an object as a class instance we will write:
:- object(Object, instantiates(Class)). ... :- end_object.
A class may specialize another class, its superclass:
:- object(Class, specializes(Superclass)). ... :- end_object.
If we are defining a reflexive system where every class is also an instance, we will probably be using the following pattern:
:- object(Class, instantiates(Metaclass), specializes(Superclass)). ... :- end_object.
In short, an object can be a stand-alone object or be part of an object hierarchy. The hierarchy can be prototype-based (defined by extending other objects) or class-based (with instantiation and specialization relations). An object may also implement one or more protocols or import one or more categories.
A stand-alone object (i.e. an object with no extension, instantiation, or specialization relations with other objects) is always compiled as a prototype, that is, a self-describing object. If we want to use classes and instances, then we will need to specify at least one instantiation or specialization relation. The best way to do this is to define a set of objects that provide the basis of a reflective system [Cointe 87, Moura 94]. For example:
:- object(object, % default root of the inheritance graph instantiates(class)). % predicates common to all objects ... :- end_object. :- object(class, % default metaclass for all classes instantiates(class), % predicates common to all instantiable classes specializes(abstract_class)). ... :- end_object. :- object(abstract_class, % default metaclass for all abstract classes instantiates(class), % predicates common to all classes specializes(object)). ... :- end_object.
Note that with these instantiation and specialization relations, object
, class
, and abstract_class
are, at the same time, classes and instances of some class. In addition, each object inherits its own predicates and the predicates of the other two objects without any inheritance loop problems.
When a full-blown reflective system solution is not needed, the above scheme can be simplified by making an object an instance of itself, i.e. by making a class its own metaclass. For example:
:- object(class, instantiates(class)). ... :- end_object.
We can use, in the same application, both prototype and class-based hierarchies (and freely exchange messages between all objects). We cannot however mix the two types of hierarchies by, e.g., specializing an object that extends another object in this current Logtalk version.
Parametric objects have a compound term for name instead of an atom. This compound term usually contains free variables that can be instantiated when sending or as a consequence of sending a message to the object, thus acting as object parameters. The object predicates can then be coded to depend on those parameters, which are logical variables shared by all object predicates. When an object state is set at object creation and never changed, parameters provide a better solution than using the object's database via asserts. Parametric objects can also be used to associate a set of predicates to terms that share a common functor and arity.
In order to give access to an object parameters, Logtalk provides the parameter/2
built-in local method:
:- object(Functor(Arg1, Arg2, ...)). ... Predicate :- ..., parameter(Number, Value), ... .
An alternative solution is to use the built-in local method this/1
. For example:
:- object(foo(Arg)). ... bar :- ..., this(foo(Arg)), ... .
Both solutions are equally efficient because the runtime cost of the methods this/1
and parameter/2
is negligible. The drawback of this second solution is that we must check all calls of this/1
if we change the object name. Note that we can't use these method with the message sending operators (::/2
, ::/1
, or ^^/1
).
When storing a parametric object in its own source file, the convention is to name the file after the object, with the object arity appended. For instance, when defining an object named sort(Type)
, we may save it in a sort_1.lgt
text file. This way it is easy to avoid file name clashes when saving Logtalk entities that have the same functor but different arity.
Compound terms with the same functor and with the same number of arguments as a parametric object identifier may act as proxies to a parametric object. Proxies may be stored on the database as Prolog facts and be used to represent different instantiations of a parametric object identifier. Logtalk provides a convenient notation for accessing proxies represented as Prolog facts when sending a message:
{Proxy}::Message
In this context, the proxy argument is proved as a plain Prolog goal. If successful, the message is sent to the corresponding parametric object. Typically, the proof allows retrieving of parameter instantiations. This construct can either be used with a proxy argument that is sufficiently instantiated in order to unify with a single Prolog fact or with a proxy argument that unifies with several facts on backtracking.
We can find, by backtracking, all defined objects by calling the current_object/1
built-in predicate with a non-instantiated variable:
| ?- current_object(Object).
This predicate can also be used to test if an object is defined by calling it with a valid object identifier (an atom or a compound term).
An object can be dynamically created at runtime by using the create_object/4
built-in predicate:
| ?- create_object(Object, Relations, Directives, Clauses).
The first argument should be either a variable or the name of the new object (a Prolog atom or compound term, which must not match any existing entity name). The remaining three arguments correspond to the relations described in the opening object directive and to the object code contents (directives and clauses).
For instance, the call:
| ?- create_object(foo, [extends(bar)], [public(foo/1)], [foo(1), foo(2)]).
is equivalent to compiling and loading the object:
:- object(foo, extends(bar)). :- dynamic. :- public(foo/1). foo(1). foo(2). :- end_object.
If we need to create a lot of (dynamic) objects at runtime, then is best to define a metaclass or a prototype with a predicate that will call this built-in predicate to make new objects. This predicate may provide automatic object name generation, name checking, and accept object initialization options.
Dynamic objects can be abolished using the abolish_object/1
built-in predicate:
| ?- abolish_object(Object).
The argument must be an identifier of a defined dynamic object, otherwise an error will be thrown.
Object directives are used to set initialization goals, define object properties, to document an object dependencies on other Logtalk entities, and to load the contents of files into an object.
We can define a goal to be executed as soon as an object is (compiled and) loaded to memory with the initialization/1
directive:
:- initialization(Goal).
The argument can be any valid Prolog or Logtalk goal, including a message to other object. For example:
:- object(foo). :- initialization(init). :- private(init/0). init :- ... . ... :- end_object.
Or:
:- object(assembler). :- initialization(control::start). ... :- end_object.
The initialization goal can also be a message to self in order to call an inherited or imported predicate. For example, assuming that we have a monitor
category defining a reset/0
predicate:
:- object(profiler, imports(monitor)). :- initialization(::reset). ... :- end_object.
Note, however, that descendant objects do not inherit initialization directives. In this context, self denotes the object that contains the directive. Also note that by initialization we do not necessarily mean setting an object dynamic state.
Similar to Prolog predicates, an object can be either static or dynamic. An object created during the execution of a program is always dynamic. An object defined in a file can be either dynamic or static. Dynamic objects are declared by using the dynamic/0
directive in the object source code:
:- dynamic.
The directive must precede any predicate directives or clauses. Please be aware that using dynamic code results in a performance hit when compared to static code. We should only use dynamic objects when these need to be abolished during program execution. In addition, note that we can declare and define dynamic predicates within a static object.
An object can be documented with arbitrary user-defined information by using the info/1
directive:
:- info(List).
See the documenting Logtalk programs section for details.
The include/1
directive can be used to load the contents of a file into an object. A typical usage scenario is to load a plain Prolog database into an object thus providing a simple way to encapsulate it. For example, assume a cities.pl
file defining facts for a city/4
predicate. We could define a wrapper for this database by writing:
:- object(cities). :- public(city/4). :- include(includes(databases('cities.pl'))). :- end_object.
The include/1
directive can also be used when creating an object dynamically. For example:
| ?- create_object(cities, [], [public(city/4), include(databases('cities.pl'))], []).
Logtalk provides six sets of built-in predicates that enable us to query the system about the possible relationships that an object may have with other entities.
The built-in predicates instantiates_class/2
and instantiates_class/3
can be used to query all instantiation relations:
| ?- instantiates_class(Instance, Class).
or, if we want to know the instantiation scope:
| ?- instantiates_class(Instance, Class, Scope).
Specialization relations can be found by using either the specializes_class/2
or the specializes_class/3
built-in predicates:
| ?- specializes_class(Class, Superclass).
or, if we want to know the specialization scope:
| ?- specializes_class(Class, Superclass, Scope).
For prototypes, we can query extension relations with the extends_object/2
or the extends_object/3
built-in predicates:
| ?- extends_object(Object, Parent).
or, if we want to know the extension scope:
| ?- extends_object(Object, Parent, Scope).
In order to find which objects import which categories we can use the built-in predicates imports_category/2
or imports_category/3
:
| ?- imports_category(Object, Category).
or, if we want to know the importation scope:
| ?- imports_category(Object, Category, Scope).
To find which objects implements which protocols we can use the implements_protocol/2-3
and conforms_to_protocol/2-3
built-in predicates:
| ?- implements_protocol(Object, Protocol, Scope).
or, if we also want inherited protocols:
| ?- conforms_to_protocol(Object, Protocol, Scope).
Note that, if we use a non-instantiated variable for the first argument, we will need to use the current_object/1
built-in predicate to ensure that the entity returned is an object and not a category.
To find which objects are explicitly complemented by categories we can use the
complements_object/2
built-in predicate:
| ?- complements_object(Category, Object).
Note that more than one category may explicitly complement a single object.
We can find the properties of defined objects by calling the built-in predicate object_property/2
:
| ?- object_property(Object, Property).
The following object properties are supported:
static
dynamic
abolish_object/1
built-in predicate)built_in
threaded
file(Path)
file(Basename, Directory)
lines(BeginLine, EndLine)
context_switching_calls
<</2
debugging control construct)dynamic_declarations
events
source_data
complements(Permission)
allow
or restrict
)complements
public(Predicates)
protected(Predicates)
private(Predicates)
declares(Predicate, Properties)
defines(Predicate, Properties)
includes(Predicate, Entity, Properties)
number_of_clauses(Number)
and line_count(Line)
with Line
being the begin line of the multifile predicate clause)provides(Predicate, Entity, Properties)
number_of_clauses(Number)
and line_count(Line)
with Line
being the begin line of the multifile predicate clause)alias(Predicate, Properties)
for(Original)
, from(Entity)
, non_terminal(NonTerminal)
, and line_count(Line)
with Line
being the begin line of the alias directive)calls(Call, Properties)
Call
is either a predicate indicator or a control construct such as ^^/1
with a predicate indicator as argument; note that Call
may not be ground when the control construct is ::/2
and the object argument is only know at runtime; the properties include caller(Caller)
and line_count(Line)
with Caller
being a predicate indicator and Line
being the begin line of the predicate clause or directive making the call)number_of_clauses(Number)
number_of_user_clauses(Number)
Some of the properties such as line numbers are only available when the object is defined in a source file compiled with the source_data
flag turned on.
Logtalk defines some built-in objects that are always available for any application.
Logtalk defines a built-in, pseudo-object named user
that virtually contains all user predicate definitions not encapsulated in a Logtalk entity. These predicates are assumed to be implicitly declared public. Messages sent from this pseudo-object, which includes messages sent from the top-level interpreter, always generate events. Defining complementing categories for this pseudo-object is not supported.
Logtalk defines a built-in object named logtalk
that provides structured message printing mechanism predicates, structured question asking predicates, debugging event predicates, predicates for accessing the internal database of loaded files and their properties, and also a set of low-level utility predicates normally used when defining hook objects.
The following predicates are defined:
expand_library_path(Library, Path)
loaded_file(Path)
loaded_file_property(Path, Property)
basename/1
, directory/1
,
flags/1
(explicit flags used when the file was loaded), text_properties/1
(list, possibly empty, whose possible elements are encoding/1
and bom/1
), target/1
(full path for the Prolog file generated by the compilation of the loaded source file), modified/1
(time stamp that should be treated as an opaque term but that may be used for comparisons), parent/1
(parent file, if it exists, that loaded the file; a file may have multiple parents), and library/1
(library name when there is a library whose location is the same as the loaded file directory).compile_aux_clauses(Clauses)
goal_expansion/2
hooks in order to compile auxiliary clauses generated for supporting an expanded goal. The compilation of the clauses avoids the risk of making the predicate whose clause is being goal-expanded discontiguous by accident.entity_prefix(Entity, Prefix)
compile_predicate_heads(Heads, Entity, TranslatedHeads, ContextArgument)
Entity
is not instantiated.compile_predicate_indicators(PredicateIndicators, Entity, TranslatedPredicateIndicators)
Entity
is not instantiated.decompile_predicate_heads(TranslatedHeads, Entity, EntityType, Heads)
decompile_predicate_indicators(TranslatedPredicateIndicators, Entity, EntityType, PredicateIndicators)
execution_context(ExecutionContext, Entity, Sender, This, Self, MetaCallContext, Stack)
print_message(Kind, Component, Term)
message_tokens//2
hook non-terminal. When the conversion fails, the message term itself is printed.print_message_tokens(Stream, Prefix, Tokens)
print_message_token(Stream, Prefix, Token, Tokens)
message_tokens(Term, Component)
message_prefix_stream(Kind, Component, Prefix, Stream)
message_hook(Term, Kind, Component, Tokens)
trace_event(Event, EventExecutionContext)
debug_handler_provider(Provider)
debug_handler(Event, EventExecutionContext)
To use these predicates, simply send the corresponding message to the logtalk
object.