The YAP Module system
======================
[TOC]
The YAP module system is based on the Quintus/SISCtus module
system ˜\cite quintus . In this design, modules are named collections of predicates,
and all predicates belong to a single module. By default, predicates are only
visible within a module, or _private_ to that module. The module
may also define a list of predicates that are
_exported_, that is, visible to other modules.
The main predicates in the module system are:
* module/2 associates a source file to a module. It has two arguments: the name of the new module, and a list of predicates exported by the module.
* use_module/1 and use_module/2 can be used to load a module. They take as first argument the source file for the module. Whereas use_module/1 loads all exported predicates, use_module/2 only takes the ones given by the second argument.
YAP pre-defines a number of modules. Most system predicates belong to
the module `prolog`. Predicates from the module `prolog` are
automatically visible to every module. The `system` module was
introduced for SWI-Prolog compatibility, and in YAP mostly acts as an
alias to `prolog`. The `user` module is also visible to all other modules.
The YAP engine is always associated to a module, the current source
module or type-in module. By default, all predicates
read-in and all calls to a goal will be made to predicates visible to
the current source module, Initially, the source module for YAP is the
module `user`. Thus Prolog programs that do not define modules will
operate within the `user` module. In this case, all predicates will be
visible to all source files.
YAP includes a number of libraries and packages, most of them
defining their own modules. Note that there is no system mechanism to
avoid clashes between module names, so it is up to the programmer to
carefully choose the names for her own program modules.
The main mechanism to change the current type-in module is by using
the module/2 declaration.This declaration sets the source module when
it starts consulting a file, and resets it at the end. One can set
the type-in module permanently by using the built-in `module/1`.
#### Explicit Naming {#ExplicitNaming}
The module system allows one to _explicitly_ specify the source mode for
a clause by prefixing a clause with its module, say:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
user:(a :- b).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
it is also possible to type
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
user:a :- user:b.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
both formulations describe the same clause, independently of the
current type-in module.
In fact, it is sufficient to specify the source mode for the clause's
head:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
user:a :- b.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
if the current type-in module is `m`, the clause could also be written as:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
user:a :- m:b.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The compiler rewrites the source clauses to ensure that explicit calls
are respected, and that implicit calls are made to the current source
module.
A goal should refer to a predicate visible within the current type-in
module. Thus, if a goal appears in a text file with a module
declaration, the goal refers to that module's context (but see the
initialization/1 directive for more details).
Again, one can override this rule by prefixing a goal with a module to
be consulted. The following query:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
?- nasa:launch(apollo,13).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
invokes the goal `launch(apollo,13)` as if the current source
module was `nasa`.
YAP and other Prolog systems allow the module prefix to see all
predicates visible in the module, including predicates private to the
module. This rule allows maximum flexibility, but it also breaks
encapsulation and should be used with care. The ciao language proposes
a different approach to this problem, see \cite DBLP:conf/cl/GrasH00 .
Modules are not always associated with a source-file. They
may range over several files, by using the
`include`directive. Moreover, they may not be associated to any source
file. As an example,
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
?- assert( nasa:launch(apollo,13) ).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
will create a module `nasa`, if does not already exist. In fact it is
sufficient to call a predicate from a module to implicitly create the
module. Hence after this call:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
?- nasa:launch(apollo,13).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
there will be a `nasa`module in the system, even if nasa:launch/2 is
not at all defined.
\pred use_module( +Files ) is directive
loads a module file
This predicate loads the file specified by _Files_, importing all
their public predicates into the current type-in module. It is
implemented as if by:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
use_module(F) :-
load_files(F, [if(not_loaded),must_be_module(true)]).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Notice that _Files_ may be a single file, or a list with a number
files. The _Files_ are loaded in YAP only once, even if they have been
updated meanwhile. YAP should also verify whether the files actually
define modules. Please consult load_files/3 for other options when
loading a file.
Predicate name clashes between two different modules may arise, either
when trying to import predicates that are also defined in the current
type-in module, or by trying to import the same predicate from two
different modules.
In the first case, the local predicate is considered to have priority
and use_module/1 simply gives a warning. As an example, if the file
`a.pl` contains:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
:- module( a, [a/1] ).
:- use_module(b).
a(1).
a(X) :- b(X).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
and the file `b.pl` contains:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
:- module( b, [a/1,b/1] ).
a(2).
b(1).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
YAP will execute as follows:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
?- [a].
% consulting .../a.pl...
% consulting .../b.pl...
% consulted .../b.pl in module b, 0 msec 0 bytes
% consulted .../a.pl in module a, 1 msec 0 bytes
true.
?- a(X).
X = 1 ? ;
X = 1.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The example shows that the query `a(X)`has a single answer, the one
defined in `a.pl`. Calls to `a(X)`succeed in the top-level, because
the module `a` was loaded into `user`. On the other hand, `b(X)`is not
exported by `a.pl`, and is not available to calls, although it can be
accessed as a predicate in the module 'a' by using the `:` operator.
Next, consider the three files `c.pl`, `d1.pl`, and `d2.pl`:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
% c.pl
:- module( c, [a/1] ).
:- use_module([d1, d2]).
a(X) :-
b(X).
a(X) :-
c(X).
a(X) :-
d(X).
% d1.pl
:- module( d1, [b/1,c/1] ).
b(2).
c(3).
% d2.pl
:- module( d2, [b/1,d/1] ).
b(1).
d(4).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The result is as follows:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
./yap -l c
YAP 6.3.4 (x86_64-darwin13.3.0): Tue Jul 15 10:42:11 CDT 2014
ERROR!!
at line 3 in o/d2.pl,
PERMISSION ERROR- loading .../c.pl: modules d1 and d2 both define b/1
?- a(X).
X = 2 ? ;
ERROR!!
EXISTENCE ERROR- procedure c/1 is undefined, called from context prolog:$user_call/2
Goal was c:c(_131290)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The state of the module system after this error is undefined.
### BuiltIn predicates {#ModuleBuiltins)
@\pred module(+ M:atom,+ L:list ) is directive
the current file defines module _M_ with exports _L_. The list may include
+ predicate indicators
+ operator definitions that look like calls to op/3.
The list _L_ may include predicates imported from other modules. If
you want to fully reexport a module, or a sub-set, also consider reexport/1.
Similar to module/2, this directive defines the file where it
appears in as a module file; it must be the first declaration in the file.
_M_ must be an atom specifying the module name; _L_ must be a
list containing the module's public predicates specification, in the
form `[predicate_name/arity,...]`.
The last argument _Options_ must be a list of options, which can be:
+ filename
the filename for a module to import into the current module.
+ library( +File )
a library file to import into the current module.
+ hide( +Opt)
if _Opt_ is `false`, keep source code for current module, if `true`, disable.
+ export(+PredicateIndicator )
Add predicates to the public list of the context module. This implies
the predicate will be imported into another module if this module
is imported with use_module/1 and use_module/2.
+ export_list(? _Mod_,? _ListOfPredicateIndicator_)
The list _ListOfPredicateIndicator_ contains all predicates
exported by module _Mod_
Note that predicates are normally exported using the directive
`module/2`. The `export/1` argumwnt is meant to allow export from
dynamically created modules. The directive argument may also be a list
of predicates.
@pred use_module(+Files, +Imports)
loads a module file but only imports the named predicates
This predicate loads the file specified by _Files_, importing their
public predicates specified by _Imports_ into the current type-in
module. It is implemented as if by:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
use_module(Files, Imports) :-
load_files(Files, [if(not_loaded),must_be_module(true),imports(Imports)]).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The _Imports_ argument may be use to specify which predicates one
wants to load. It can also be used to give the predicates a different name. As an example,
the graphs library is implemented on top of the red-black trees library, and some predicates are just aliases:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~{.prolog}
:- use_module(library(rbtrees), [
rb_min/3 as min_assoc,
rb_max/3 as max_assoc,
...]).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Unfortunately it is still not possible to change argument order.
\pred module(+ M:atom,+ L:list ) is directive
the current file defines module _M_ with exports _L_. The list may include
+ predicate indicators
+ operator definitions that look like calls to op/3.
The list _L_ may include predicates imported from other modules. If
you want to fully reexport a module, or a sub-set, also consider reexport/1.
Similar to module/2, this directive defines the file where it
appears in as a module file; it must be the first declaration in the file.
_M_ must be an atom specifying the module name; _L_ must be a
list containing the module's public predicates specification, in the
form `[predicate_name/arity,...]`.
The last argument _Options_ must be a list of options, which can be:
+filename
the filename for a module to import into the current module.
+ library( +File )
a library file to import into the current module.
+ hide( +Opt)
if _Opt_ is `false`, keep source code for current module, if `true`, disable.
+ export(+PredicateIndicator )
Add predicates to the public list of the context module. This implies
the predicate will be imported into another module if this module
is imported with use_module/1 and use_module/2.
+ export_list(? _Mod_,? _ListOfPredicateIndicator_)
The list _ListOfPredicateIndicator_ contains all predicates
exported by module _Mod_
Note that predicates are normally exported using the directive
`module/2`. The `export/1` argument is meant to allow export from
dynamically created modules. The directive argument may also be a list
of predicates.