9 $ {$package . "::OVERLOAD"}{dummy}++; # Register with magic by touching.
10 *{$package . "::()"} = \&nil; # Make it findable via fetchmethod.
12 if ($_ eq 'fallback') {
16 if (not ref $sub and $sub !~ /::/) {
17 $ {$package . "::(" . $_} = $sub;
20 #print STDERR "Setting `$ {'package'}::\cO$_' to \\&`$sub'.\n";
21 *{$package . "::(" . $_} = \&{ $sub };
24 ${$package . "::()"} = $fb; # Make it findable too (fallback only).
28 $package = (caller())[0];
29 # *{$package . "::OVERLOAD"} = \&OVERLOAD;
31 $package->overload::OVERLOAD(@_);
35 $package = (caller())[0];
36 ${$package . "::OVERLOAD"}{dummy}++; # Upgrade the table
39 if ($_ eq 'fallback') {
40 undef $ {$package . "::()"};
42 delete $ {$package . "::"}{"(" . $_};
49 $package = ref $package if ref $package;
55 return undef unless $globref;
56 my $sub = \&{*$globref};
57 return $sub if $sub ne \&nil;
58 return shift->can($ {*$globref});
61 sub OverloadedStringify {
63 $package = ref $package if ref $package;
65 ov_method mycan($package, '(""'), $package
66 or ov_method mycan($package, '(0+'), $package
67 or ov_method mycan($package, '(bool'), $package
68 or ov_method mycan($package, '(nomethod'), $package;
73 $package = ref $package if ref $package;
74 #my $meth = $package->can('(' . shift);
75 ov_method mycan($package, '(' . shift), $package;
76 #return $meth if $meth ne \&nil;
81 my $package = ref $_[0];
82 return "$_[0]" unless $package;
83 bless $_[0], overload::Fake; # Non-overloaded package
85 bless $_[0], $package; # Back
86 $package . substr $str, index $str, '=';
90 (OverloadedStringify($_[0]) or ref($_[0]) eq 'Regexp') ?
95 sub mycan { # Real can would leave stubs.
96 my ($package, $meth) = @_;
97 return \*{$package . "::$meth"} if defined &{$package . "::$meth"};
99 foreach $p (@{$package . "::ISA"}) {
100 my $out = mycan($p, $meth);
114 %ops = ( with_assign => "+ - * / % ** << >> x .",
115 assign => "+= -= *= /= %= **= <<= >>= x= .=",
116 num_comparison => "< <= > >= == !=",
117 '3way_comparison'=> "<=> cmp",
118 str_comparison => "lt le gt ge eq ne",
122 func => "atan2 cos sin exp abs log sqrt",
123 conversion => 'bool "" 0+',
125 dereferencing => '${} @{} %{} &{} *{}',
126 special => 'nomethod fallback =');
129 # Arguments: what, sub
132 $^H |= $constants{$_[0]} | 0x20000;
137 sub remove_constant {
138 # Arguments: what, sub
141 $^H &= ~ $constants{$_[0]};
152 overload - Package for overloading perl operations
165 $a = new SomeThing 57;
168 if (overload::Overloaded $b) {...}
170 $strval = overload::StrVal $b;
174 =head2 Declaration of overloaded functions
176 The compilation directive
183 declares function Number::add() for addition, and method muas() in
184 the "class" C<Number> (or one of its base classes)
185 for the assignment form C<*=> of multiplication.
187 Arguments of this directive come in (key, value) pairs. Legal values
188 are values legal inside a C<&{ ... }> call, so the name of a
189 subroutine, a reference to a subroutine, or an anonymous subroutine
190 will all work. Note that values specified as strings are
191 interpreted as methods, not subroutines. Legal keys are listed below.
193 The subroutine C<add> will be called to execute C<$a+$b> if $a
194 is a reference to an object blessed into the package C<Number>, or if $a is
195 not an object from a package with defined mathemagic addition, but $b is a
196 reference to a C<Number>. It can also be called in other situations, like
197 C<$a+=7>, or C<$a++>. See L<MAGIC AUTOGENERATION>. (Mathemagical
198 methods refer to methods triggered by an overloaded mathematical
201 Since overloading respects inheritance via the @ISA hierarchy, the
202 above declaration would also trigger overloading of C<+> and C<*=> in
203 all the packages which inherit from C<Number>.
205 =head2 Calling Conventions for Binary Operations
207 The functions specified in the C<use overload ...> directive are called
208 with three (in one particular case with four, see L<Last Resort>)
209 arguments. If the corresponding operation is binary, then the first
210 two arguments are the two arguments of the operation. However, due to
211 general object calling conventions, the first argument should always be
212 an object in the package, so in the situation of C<7+$a>, the
213 order of the arguments is interchanged. It probably does not matter
214 when implementing the addition method, but whether the arguments
215 are reversed is vital to the subtraction method. The method can
216 query this information by examining the third argument, which can take
217 three different values:
223 the order of arguments is as in the current operation.
227 the arguments are reversed.
231 the current operation is an assignment variant (as in
232 C<$a+=7>), but the usual function is called instead. This additional
233 information can be used to generate some optimizations. Compare
234 L<Calling Conventions for Mutators>.
238 =head2 Calling Conventions for Unary Operations
240 Unary operation are considered binary operations with the second
241 argument being C<undef>. Thus the functions that overloads C<{"++"}>
242 is called with arguments C<($a,undef,'')> when $a++ is executed.
244 =head2 Calling Conventions for Mutators
246 Two types of mutators have different calling conventions:
250 =item C<++> and C<-->
252 The routines which implement these operators are expected to actually
253 I<mutate> their arguments. So, assuming that $obj is a reference to a
256 sub incr { my $n = $ {$_[0]}; ++$n; $_[0] = bless \$n}
258 is an appropriate implementation of overloaded C<++>. Note that
260 sub incr { ++$ {$_[0]} ; shift }
262 is OK if used with preincrement and with postincrement. (In the case
263 of postincrement a copying will be performed, see L<Copy Constructor>.)
265 =item C<x=> and other assignment versions
267 There is nothing special about these methods. They may change the
268 value of their arguments, and may leave it as is. The result is going
269 to be assigned to the value in the left-hand-side if different from
272 This allows for the same method to be used as overloaded C<+=> and
273 C<+>. Note that this is I<allowed>, but not recommended, since by the
274 semantic of L<"Fallback"> Perl will call the method for C<+> anyway,
275 if C<+=> is not overloaded.
279 B<Warning.> Due to the presense of assignment versions of operations,
280 routines which may be called in assignment context may create
281 self-referential structures. Currently Perl will not free self-referential
282 structures until cycles are C<explicitly> broken. You may get problems
283 when traversing your structures too.
287 use overload '+' => sub { bless [ \$_[0], \$_[1] ] };
289 is asking for trouble, since for code C<$obj += $foo> the subroutine
290 is called as C<$obj = add($obj, $foo, undef)>, or C<$obj = [\$obj,
291 \$foo]>. If using such a subroutine is an important optimization, one
292 can overload C<+=> explicitly by a non-"optimized" version, or switch
293 to non-optimized version if C<not defined $_[2]> (see
294 L<Calling Conventions for Binary Operations>).
296 Even if no I<explicit> assignment-variants of operators are present in
297 the script, they may be generated by the optimizer. Say, C<",$obj,"> or
298 C<',' . $obj . ','> may be both optimized to
300 my $tmp = ',' . $obj; $tmp .= ',';
302 =head2 Overloadable Operations
304 The following symbols can be specified in C<use overload> directive:
308 =item * I<Arithmetic operations>
310 "+", "+=", "-", "-=", "*", "*=", "/", "/=", "%", "%=",
311 "**", "**=", "<<", "<<=", ">>", ">>=", "x", "x=", ".", ".=",
313 For these operations a substituted non-assignment variant can be called if
314 the assignment variant is not available. Methods for operations "C<+>",
315 "C<->", "C<+=>", and "C<-=>" can be called to automatically generate
316 increment and decrement methods. The operation "C<->" can be used to
317 autogenerate missing methods for unary minus or C<abs>.
319 See L<"MAGIC AUTOGENERATION">, L<"Calling Conventions for Mutators"> and
320 L<"Calling Conventions for Binary Operations">) for details of these
323 =item * I<Comparison operations>
325 "<", "<=", ">", ">=", "==", "!=", "<=>",
326 "lt", "le", "gt", "ge", "eq", "ne", "cmp",
328 If the corresponding "spaceship" variant is available, it can be
329 used to substitute for the missing operation. During C<sort>ing
330 arrays, C<cmp> is used to compare values subject to C<use overload>.
332 =item * I<Bit operations>
334 "&", "^", "|", "neg", "!", "~",
336 "C<neg>" stands for unary minus. If the method for C<neg> is not
337 specified, it can be autogenerated using the method for
338 subtraction. If the method for "C<!>" is not specified, it can be
339 autogenerated using the methods for "C<bool>", or "C<\"\">", or "C<0+>".
341 =item * I<Increment and decrement>
345 If undefined, addition and subtraction methods can be
346 used instead. These operations are called both in prefix and
349 =item * I<Transcendental functions>
351 "atan2", "cos", "sin", "exp", "abs", "log", "sqrt",
353 If C<abs> is unavailable, it can be autogenerated using methods
354 for "E<lt>" or "E<lt>=E<gt>" combined with either unary minus or subtraction.
356 =item * I<Boolean, string and numeric conversion>
358 "bool", "\"\"", "0+",
360 If one or two of these operations are not overloaded, the remaining ones can
361 be used instead. C<bool> is used in the flow control operators
362 (like C<while>) and for the ternary "C<?:>" operation. These functions can
363 return any arbitrary Perl value. If the corresponding operation for this value
364 is overloaded too, that operation will be called again with this value.
370 If not overloaded, the argument will be converted to a filehandle or
371 glob (which may require a stringification). The same overloading
372 happens both for the I<read-filehandle> syntax C<E<lt>$varE<gt>> and
373 I<globbing> syntax C<E<lt>${var}E<gt>>.
375 =item * I<Dereferencing>
377 '${}', '@{}', '%{}', '&{}', '*{}'.
379 If not overloaded, the argument will be dereferenced I<as is>, thus
380 should be of correct type. These functions should return a reference
381 of correct type, or another object with overloaded dereferencing.
385 "nomethod", "fallback", "=",
387 see L<SPECIAL SYMBOLS FOR C<use overload>>.
391 See L<"Fallback"> for an explanation of when a missing method can be
394 A computer-readable form of the above table is available in the hash
395 %overload::ops, with values being space-separated lists of names:
397 with_assign => '+ - * / % ** << >> x .',
398 assign => '+= -= *= /= %= **= <<= >>= x= .=',
399 num_comparison => '< <= > >= == !=',
400 '3way_comparison'=> '<=> cmp',
401 str_comparison => 'lt le gt ge eq ne',
405 func => 'atan2 cos sin exp abs log sqrt',
406 conversion => 'bool "" 0+',
408 dereferencing => '${} @{} %{} &{} *{}',
409 special => 'nomethod fallback ='
411 =head2 Inheritance and overloading
413 Inheritance interacts with overloading in two ways.
417 =item Strings as values of C<use overload> directive
421 use overload key => value;
423 is a string, it is interpreted as a method name.
425 =item Overloading of an operation is inherited by derived classes
427 Any class derived from an overloaded class is also overloaded. The
428 set of overloaded methods is the union of overloaded methods of all
429 the ancestors. If some method is overloaded in several ancestor, then
430 which description will be used is decided by the usual inheritance
433 If C<A> inherits from C<B> and C<C> (in this order), C<B> overloads
434 C<+> with C<\&D::plus_sub>, and C<C> overloads C<+> by C<"plus_meth">,
435 then the subroutine C<D::plus_sub> will be called to implement
436 operation C<+> for an object in package C<A>.
440 Note that since the value of the C<fallback> key is not a subroutine,
441 its inheritance is not governed by the above rules. In the current
442 implementation, the value of C<fallback> in the first overloaded
443 ancestor is used, but this is accidental and subject to change.
445 =head1 SPECIAL SYMBOLS FOR C<use overload>
447 Three keys are recognized by Perl that are not covered by the above
452 C<"nomethod"> should be followed by a reference to a function of four
453 parameters. If defined, it is called when the overloading mechanism
454 cannot find a method for some operation. The first three arguments of
455 this function coincide with the arguments for the corresponding method if
456 it were found, the fourth argument is the symbol
457 corresponding to the missing method. If several methods are tried,
458 the last one is used. Say, C<1-$a> can be equivalent to
460 &nomethodMethod($a,1,1,"-")
462 if the pair C<"nomethod" =E<gt> "nomethodMethod"> was specified in the
463 C<use overload> directive.
465 If some operation cannot be resolved, and there is no function
466 assigned to C<"nomethod">, then an exception will be raised via die()--
467 unless C<"fallback"> was specified as a key in C<use overload> directive.
471 The key C<"fallback"> governs what to do if a method for a particular
472 operation is not found. Three different cases are possible depending on
473 the value of C<"fallback">:
480 substituted method (see L<MAGIC AUTOGENERATION>). If this fails, it
481 then tries to calls C<"nomethod"> value; if missing, an exception
486 The same as for the C<undef> value, but no exception is raised. Instead,
487 it silently reverts to what it would have done were there no C<use overload>
490 =item * defined, but FALSE
492 No autogeneration is tried. Perl tries to call
493 C<"nomethod"> value, and if this is missing, raises an exception.
497 B<Note.> C<"fallback"> inheritance via @ISA is not carved in stone
498 yet, see L<"Inheritance and overloading">.
500 =head2 Copy Constructor
502 The value for C<"="> is a reference to a function with three
503 arguments, i.e., it looks like the other values in C<use
504 overload>. However, it does not overload the Perl assignment
505 operator. This would go against Camel hair.
507 This operation is called in the situations when a mutator is applied
508 to a reference that shares its object with some other reference, such
514 To make this change $a and not change $b, a copy of C<$$a> is made,
515 and $a is assigned a reference to this new object. This operation is
516 done during execution of the C<++$a>, and not during the assignment,
517 (so before the increment C<$$a> coincides with C<$$b>). This is only
518 done if C<++> is expressed via a method for C<'++'> or C<'+='> (or
519 C<nomethod>). Note that if this operation is expressed via C<'+'>
520 a nonmutator, i.e., as in
525 then C<$a> does not reference a new copy of C<$$a>, since $$a does not
526 appear as lvalue when the above code is executed.
528 If the copy constructor is required during the execution of some mutator,
529 but a method for C<'='> was not specified, it can be autogenerated as a
530 string copy if the object is a plain scalar.
536 The actually executed code for
539 Something else which does not modify $a or $b....
545 Something else which does not modify $a or $b....
546 $a = $a->clone(undef,"");
549 if $b was mathemagical, and C<'++'> was overloaded with C<\&incr>,
550 C<'='> was overloaded with C<\&clone>.
554 Same behaviour is triggered by C<$b = $a++>, which is consider a synonym for
557 =head1 MAGIC AUTOGENERATION
559 If a method for an operation is not found, and the value for C<"fallback"> is
560 TRUE or undefined, Perl tries to autogenerate a substitute method for
561 the missing operation based on the defined operations. Autogenerated method
562 substitutions are possible for the following operations:
566 =item I<Assignment forms of arithmetic operations>
568 C<$a+=$b> can use the method for C<"+"> if the method for C<"+=">
571 =item I<Conversion operations>
573 String, numeric, and boolean conversion are calculated in terms of one
574 another if not all of them are defined.
576 =item I<Increment and decrement>
578 The C<++$a> operation can be expressed in terms of C<$a+=1> or C<$a+1>,
579 and C<$a--> in terms of C<$a-=1> and C<$a-1>.
583 can be expressed in terms of C<$aE<lt>0> and C<-$a> (or C<0-$a>).
587 can be expressed in terms of subtraction.
591 C<!> and C<not> can be expressed in terms of boolean conversion, or
592 string or numerical conversion.
594 =item I<Concatenation>
596 can be expressed in terms of string conversion.
598 =item I<Comparison operations>
600 can be expressed in terms of its "spaceship" counterpart: either
601 C<E<lt>=E<gt>> or C<cmp>:
603 <, >, <=, >=, ==, != in terms of <=>
604 lt, gt, le, ge, eq, ne in terms of cmp
608 <> in terms of builtin operations
610 =item I<Dereferencing>
612 ${} @{} %{} &{} *{} in terms of builtin operations
614 =item I<Copy operator>
616 can be expressed in terms of an assignment to the dereferenced value, if this
617 value is a scalar and not a reference.
621 =head1 Losing overloading
623 The restriction for the comparison operation is that even if, for example,
624 `C<cmp>' should return a blessed reference, the autogenerated `C<lt>'
625 function will produce only a standard logical value based on the
626 numerical value of the result of `C<cmp>'. In particular, a working
627 numeric conversion is needed in this case (possibly expressed in terms of
630 Similarly, C<.=> and C<x=> operators lose their mathemagical properties
631 if the string conversion substitution is applied.
633 When you chop() a mathemagical object it is promoted to a string and its
634 mathemagical properties are lost. The same can happen with other
637 =head1 Run-time Overloading
639 Since all C<use> directives are executed at compile-time, the only way to
640 change overloading during run-time is to
642 eval 'use overload "+" => \&addmethod';
646 eval 'no overload "+", "--", "<="';
648 though the use of these constructs during run-time is questionable.
650 =head1 Public functions
652 Package C<overload.pm> provides the following public functions:
656 =item overload::StrVal(arg)
658 Gives string value of C<arg> as in absence of stringify overloading.
660 =item overload::Overloaded(arg)
662 Returns true if C<arg> is subject to overloading of some operations.
664 =item overload::Method(obj,op)
666 Returns C<undef> or a reference to the method that implements C<op>.
670 =head1 Overloading constants
672 For some application Perl parser mangles constants too much. It is possible
673 to hook into this process via overload::constant() and overload::remove_constant()
676 These functions take a hash as an argument. The recognized keys of this hash
683 to overload integer constants,
687 to overload floating point constants,
691 to overload octal and hexadecimal constants,
695 to overload C<q>-quoted strings, constant pieces of C<qq>- and C<qx>-quoted
696 strings and here-documents,
700 to overload constant pieces of regular expressions.
704 The corresponding values are references to functions which take three arguments:
705 the first one is the I<initial> string form of the constant, the second one
706 is how Perl interprets this constant, the third one is how the constant is used.
707 Note that the initial string form does not
708 contain string delimiters, and has backslashes in backslash-delimiter
709 combinations stripped (thus the value of delimiter is not relevant for
710 processing of this string). The return value of this function is how this
711 constant is going to be interpreted by Perl. The third argument is undefined
712 unless for overloaded C<q>- and C<qr>- constants, it is C<q> in single-quote
713 context (comes from strings, regular expressions, and single-quote HERE
714 documents), it is C<tr> for arguments of C<tr>/C<y> operators,
715 it is C<s> for right-hand side of C<s>-operator, and it is C<qq> otherwise.
717 Since an expression C<"ab$cd,,"> is just a shortcut for C<'ab' . $cd . ',,'>,
718 it is expected that overloaded constant strings are equipped with reasonable
719 overloaded catenation operator, otherwise absurd results will result.
720 Similarly, negative numbers are considered as negations of positive constants.
722 Note that it is probably meaningless to call the functions overload::constant()
723 and overload::remove_constant() from anywhere but import() and unimport() methods.
724 From these methods they may be called as
729 die "unknown import: @_" unless @_ == 1 and $_[0] eq ':constant';
730 overload::constant integer => sub {Math::BigInt->new(shift)};
733 B<BUGS> Currently overloaded-ness of constants does not propagate
736 =head1 IMPLEMENTATION
738 What follows is subject to change RSN.
740 The table of methods for all operations is cached in magic for the
741 symbol table hash for the package. The cache is invalidated during
742 processing of C<use overload>, C<no overload>, new function
743 definitions, and changes in @ISA. However, this invalidation remains
744 unprocessed until the next C<bless>ing into the package. Hence if you
745 want to change overloading structure dynamically, you'll need an
746 additional (fake) C<bless>ing to update the table.
748 (Every SVish thing has a magic queue, and magic is an entry in that
749 queue. This is how a single variable may participate in multiple
750 forms of magic simultaneously. For instance, environment variables
751 regularly have two forms at once: their %ENV magic and their taint
752 magic. However, the magic which implements overloading is applied to
753 the stashes, which are rarely used directly, thus should not slow down
756 If an object belongs to a package using overload, it carries a special
757 flag. Thus the only speed penalty during arithmetic operations without
758 overloading is the checking of this flag.
760 In fact, if C<use overload> is not present, there is almost no overhead
761 for overloadable operations, so most programs should not suffer
762 measurable performance penalties. A considerable effort was made to
763 minimize the overhead when overload is used in some package, but the
764 arguments in question do not belong to packages using overload. When
765 in doubt, test your speed with C<use overload> and without it. So far
766 there have been no reports of substantial speed degradation if Perl is
767 compiled with optimization turned on.
769 There is no size penalty for data if overload is not used. The only
770 size penalty if overload is used in some package is that I<all> the
771 packages acquire a magic during the next C<bless>ing into the
772 package. This magic is three-words-long for packages without
773 overloading, and carries the cache table if the package is overloaded.
775 Copying (C<$a=$b>) is shallow; however, a one-level-deep copying is
776 carried out before any operation that can imply an assignment to the
777 object $a (or $b) refers to, like C<$a++>. You can override this
778 behavior by defining your own copy constructor (see L<"Copy Constructor">).
780 It is expected that arguments to methods that are not explicitly supposed
781 to be changed are constant (but this is not enforced).
783 =head1 Metaphor clash
785 One may wonder why the semantic of overloaded C<=> is so counter intuitive.
786 If it I<looks> counter intuitive to you, you are subject to a metaphor
789 Here is a Perl object metaphor:
791 I< object is a reference to blessed data>
793 and an arithmetic metaphor:
795 I< object is a thing by itself>.
797 The I<main> problem of overloading C<=> is the fact that these metaphors
798 imply different actions on the assignment C<$a = $b> if $a and $b are
799 objects. Perl-think implies that $a becomes a reference to whatever
800 $b was referencing. Arithmetic-think implies that the value of "object"
801 $a is changed to become the value of the object $b, preserving the fact
802 that $a and $b are separate entities.
804 The difference is not relevant in the absence of mutators. After
805 a Perl-way assignment an operation which mutates the data referenced by $a
806 would change the data referenced by $b too. Effectively, after
807 C<$a = $b> values of $a and $b become I<indistinguishable>.
809 On the other hand, anyone who has used algebraic notation knows the
810 expressive power of the arithmetic metaphor. Overloading works hard
811 to enable this metaphor while preserving the Perlian way as far as
812 possible. Since it is not not possible to freely mix two contradicting
813 metaphors, overloading allows the arithmetic way to write things I<as
814 far as all the mutators are called via overloaded access only>. The
815 way it is done is described in L<Copy Constructor>.
817 If some mutator methods are directly applied to the overloaded values,
818 one may need to I<explicitly unlink> other values which references the
823 $b = $a; # $b is "linked" to $a
825 $a = $a->clone; # Unlink $b from $a
828 Note that overloaded access makes this transparent:
831 $b = $a; # $b is "linked" to $a
832 $a += 4; # would unlink $b automagically
834 However, it would not make
837 $a = 4; # Now $a is a plain 4, not 'Data'
839 preserve "objectness" of $a. But Perl I<has> a way to make assignments
840 to an object do whatever you want. It is just not the overload, but
841 tie()ing interface (see L<perlfunc/tie>). Adding a FETCH() method
842 which returns the object itself, and STORE() method which changes the
843 value of the object, one can reproduce the arithmetic metaphor in its
844 completeness, at least for variables which were tie()d from the start.
846 (Note that a workaround for a bug may be needed, see L<"BUGS">.)
850 Please add examples to what follows!
852 =head2 Two-face scalars
854 Put this in F<two_face.pm> in your Perl library directory:
856 package two_face; # Scalars with separate string and
858 sub new { my $p = shift; bless [@_], $p }
859 use overload '""' => \&str, '0+' => \&num, fallback => 1;
866 my $seven = new two_face ("vii", 7);
867 printf "seven=$seven, seven=%d, eight=%d\n", $seven, $seven+1;
868 print "seven contains `i'\n" if $seven =~ /i/;
870 (The second line creates a scalar which has both a string value, and a
871 numeric value.) This prints:
873 seven=vii, seven=7, eight=8
876 =head2 Two-face references
878 Suppose you want to create an object which is accessible as both an
879 array reference, and a hash reference, similar to the builtin
880 L<array-accessible-as-a-hash|perlref/"Pseudo-hashes: Using an array as
881 a hash"> builtin Perl type. Let us make it better than the builtin
882 type, there will be no restriction that you cannot use the index 0 of
886 use overload '%{}' => \&gethash, '@{}' => sub { $ {shift()} };
894 tie %h, ref $self, $self;
898 sub TIEHASH { my $p = shift; bless \ shift, $p }
901 $fields{$_} = $i++ foreach qw{zero one two three};
903 my $self = ${shift()};
904 my $key = $fields{shift()};
905 defined $key or die "Out of band access";
906 $$self->[$key] = shift;
909 my $self = ${shift()};
910 my $key = $fields{shift()};
911 defined $key or die "Out of band access";
915 Now one can access an object using both the array and hash syntax:
917 my $bar = new two_refs 3,4,5,6;
919 $bar->{two} == 11 or die 'bad hash fetch';
921 Note several important features of this example. First of all, the
922 I<actual> type of $bar is a scalar reference, and we do not overload
923 the scalar dereference. Thus we can get the I<actual> non-overloaded
924 contents of $bar by just using C<$$bar> (what we do in functions which
925 overload dereference). Similarly, the object returned by the
926 TIEHASH() method is a scalar reference.
928 Second, we create a new tied hash each time the hash syntax is used.
929 This allows us not to worry about a possibility of a reference loop,
930 would would lead to a memory leak.
932 Both these problems can be cured. Say, if we want to overload hash
933 dereference on a reference to an object which is I<implemented> as a
934 hash itself, the only problem one has to circumvent is how to access
935 this I<actual> hash (as opposed to the I<virtual> exhibited by
936 overloaded dereference operator). Here is one possible fetching routine:
939 my ($self, $key) = (shift, shift);
940 my $class = ref $self;
941 bless $self, 'overload::dummy'; # Disable overloading of %{}
942 my $out = $self->{$key};
943 bless $self, $class; # Restore overloading
947 To move creation of the tied hash on each access, one may an extra
948 level of indirection which allows a non-circular structure of references:
951 use overload '%{}' => sub { ${shift()}->[1] },
952 '@{}' => sub { ${shift()}->[0] };
958 bless \ [$a, \%h], $p;
963 tie %h, ref $self, $self;
967 sub TIEHASH { my $p = shift; bless \ shift, $p }
970 $fields{$_} = $i++ foreach qw{zero one two three};
973 my $key = $fields{shift()};
974 defined $key or die "Out of band access";
979 my $key = $fields{shift()};
980 defined $key or die "Out of band access";
984 Now if $baz is overloaded like this, then C<$bar> is a reference to a
985 reference to the intermediate array, which keeps a reference to an
986 actual array, and the access hash. The tie()ing object for the access
987 hash is also a reference to a reference to the actual array, so
993 There are no loops of references.
997 Both "objects" which are blessed into the class C<two_refs1> are
998 references to a reference to an array, thus references to a I<scalar>.
999 Thus the accessor expression C<$$foo-E<gt>[$ind]> involves no
1000 overloaded operations.
1004 =head2 Symbolic calculator
1006 Put this in F<symbolic.pm> in your Perl library directory:
1008 package symbolic; # Primitive symbolic calculator
1009 use overload nomethod => \&wrap;
1011 sub new { shift; bless ['n', @_] }
1013 my ($obj, $other, $inv, $meth) = @_;
1014 ($obj, $other) = ($other, $obj) if $inv;
1015 bless [$meth, $obj, $other];
1018 This module is very unusual as overloaded modules go: it does not
1019 provide any usual overloaded operators, instead it provides the L<Last
1020 Resort> operator C<nomethod>. In this example the corresponding
1021 subroutine returns an object which encapsulates operations done over
1022 the objects: C<new symbolic 3> contains C<['n', 3]>, C<2 + new
1023 symbolic 3> contains C<['+', 2, ['n', 3]]>.
1025 Here is an example of the script which "calculates" the side of
1026 circumscribed octagon using the above package:
1029 my $iter = 1; # 2**($iter+2) = 8
1030 my $side = new symbolic 1;
1034 $side = (sqrt(1 + $side**2) - 1)/$side;
1038 The value of $side is
1040 ['/', ['-', ['sqrt', ['+', 1, ['**', ['n', 1], 2]],
1041 undef], 1], ['n', 1]]
1043 Note that while we obtained this value using a nice little script,
1044 there is no simple way to I<use> this value. In fact this value may
1045 be inspected in debugger (see L<perldebug>), but ony if
1046 C<bareStringify> B<O>ption is set, and not via C<p> command.
1048 If one attempts to print this value, then the overloaded operator
1049 C<""> will be called, which will call C<nomethod> operator. The
1050 result of this operator will be stringified again, but this result is
1051 again of type C<symbolic>, which will lead to an infinite loop.
1053 Add a pretty-printer method to the module F<symbolic.pm>:
1056 my ($meth, $a, $b) = @{+shift};
1057 $a = 'u' unless defined $a;
1058 $b = 'u' unless defined $b;
1059 $a = $a->pretty if ref $a;
1060 $b = $b->pretty if ref $b;
1064 Now one can finish the script by
1066 print "side = ", $side->pretty, "\n";
1068 The method C<pretty> is doing object-to-string conversion, so it
1069 is natural to overload the operator C<""> using this method. However,
1070 inside such a method it is not necessary to pretty-print the
1071 I<components> $a and $b of an object. In the above subroutine
1072 C<"[$meth $a $b]"> is a catenation of some strings and components $a
1073 and $b. If these components use overloading, the catenation operator
1074 will look for an overloaded operator C<.>, if not present, it will
1075 look for an overloaded operator C<"">. Thus it is enough to use
1077 use overload nomethod => \&wrap, '""' => \&str;
1079 my ($meth, $a, $b) = @{+shift};
1080 $a = 'u' unless defined $a;
1081 $b = 'u' unless defined $b;
1085 Now one can change the last line of the script to
1087 print "side = $side\n";
1091 side = [/ [- [sqrt [+ 1 [** [n 1 u] 2]] u] 1] [n 1 u]]
1093 and one can inspect the value in debugger using all the possible
1096 Something is is still amiss: consider the loop variable $cnt of the
1097 script. It was a number, not an object. We cannot make this value of
1098 type C<symbolic>, since then the loop will not terminate.
1100 Indeed, to terminate the cycle, the $cnt should become false.
1101 However, the operator C<bool> for checking falsity is overloaded (this
1102 time via overloaded C<"">), and returns a long string, thus any object
1103 of type C<symbolic> is true. To overcome this, we need a way to
1104 compare an object to 0. In fact, it is easier to write a numeric
1107 Here is the text of F<symbolic.pm> with such a routine added (and
1108 slightly modified str()):
1110 package symbolic; # Primitive symbolic calculator
1112 nomethod => \&wrap, '""' => \&str, '0+' => \#
1114 sub new { shift; bless ['n', @_] }
1116 my ($obj, $other, $inv, $meth) = @_;
1117 ($obj, $other) = ($other, $obj) if $inv;
1118 bless [$meth, $obj, $other];
1121 my ($meth, $a, $b) = @{+shift};
1122 $a = 'u' unless defined $a;
1129 my %subr = ( n => sub {$_[0]},
1130 sqrt => sub {sqrt $_[0]},
1131 '-' => sub {shift() - shift()},
1132 '+' => sub {shift() + shift()},
1133 '/' => sub {shift() / shift()},
1134 '*' => sub {shift() * shift()},
1135 '**' => sub {shift() ** shift()},
1138 my ($meth, $a, $b) = @{+shift};
1139 my $subr = $subr{$meth}
1140 or die "Do not know how to ($meth) in symbolic";
1141 $a = $a->num if ref $a eq __PACKAGE__;
1142 $b = $b->num if ref $b eq __PACKAGE__;
1146 All the work of numeric conversion is done in %subr and num(). Of
1147 course, %subr is not complete, it contains only operators used in the
1148 example below. Here is the extra-credit question: why do we need an
1149 explicit recursion in num()? (Answer is at the end of this section.)
1151 Use this module like this:
1154 my $iter = new symbolic 2; # 16-gon
1155 my $side = new symbolic 1;
1159 $cnt = $cnt - 1; # Mutator `--' not implemented
1160 $side = (sqrt(1 + $side**2) - 1)/$side;
1162 printf "%s=%f\n", $side, $side;
1163 printf "pi=%f\n", $side*(2**($iter+2));
1165 It prints (without so many line breaks)
1167 [/ [- [sqrt [+ 1 [** [/ [- [sqrt [+ 1 [** [n 1] 2]]] 1]
1169 [/ [- [sqrt [+ 1 [** [n 1] 2]]] 1] [n 1]]]=0.198912
1172 The above module is very primitive. It does not implement
1173 mutator methods (C<++>, C<-=> and so on), does not do deep copying
1174 (not required without mutators!), and implements only those arithmetic
1175 operations which are used in the example.
1177 To implement most arithmetic operations is easy, one should just use
1178 the tables of operations, and change the code which fills %subr to
1180 my %subr = ( 'n' => sub {$_[0]} );
1181 foreach my $op (split " ", $overload::ops{with_assign}) {
1182 $subr{$op} = $subr{"$op="} = eval "sub {shift() $op shift()}";
1184 my @bins = qw(binary 3way_comparison num_comparison str_comparison);
1185 foreach my $op (split " ", "@overload::ops{ @bins }") {
1186 $subr{$op} = eval "sub {shift() $op shift()}";
1188 foreach my $op (split " ", "@overload::ops{qw(unary func)}") {
1189 print "defining `$op'\n";
1190 $subr{$op} = eval "sub {$op shift()}";
1193 Due to L<Calling Conventions for Mutators>, we do not need anything
1194 special to make C<+=> and friends work, except filling C<+=> entry of
1195 %subr, and defining a copy constructor (needed since Perl has no
1196 way to know that the implementation of C<'+='> does not mutate
1197 the argument, compare L<Copy Constructor>).
1199 To implement a copy constructor, add C<'=' => \&cpy> to C<use overload>
1200 line, and code (this code assumes that mutators change things one level
1201 deep only, so recursive copying is not needed):
1205 bless [@$self], ref $self;
1208 To make C<++> and C<--> work, we need to implement actual mutators,
1209 either directly, or in C<nomethod>. We continue to do things inside
1210 C<nomethod>, thus add
1212 if ($meth eq '++' or $meth eq '--') {
1213 @$obj = ($meth, (bless [@$obj]), 1); # Avoid circular reference
1217 after the first line of wrap(). This is not a most effective
1218 implementation, one may consider
1220 sub inc { $_[0] = bless ['++', shift, 1]; }
1224 As a final remark, note that one can fill %subr by
1226 my %subr = ( 'n' => sub {$_[0]} );
1227 foreach my $op (split " ", $overload::ops{with_assign}) {
1228 $subr{$op} = $subr{"$op="} = eval "sub {shift() $op shift()}";
1230 my @bins = qw(binary 3way_comparison num_comparison str_comparison);
1231 foreach my $op (split " ", "@overload::ops{ @bins }") {
1232 $subr{$op} = eval "sub {shift() $op shift()}";
1234 foreach my $op (split " ", "@overload::ops{qw(unary func)}") {
1235 $subr{$op} = eval "sub {$op shift()}";
1237 $subr{'++'} = $subr{'+'};
1238 $subr{'--'} = $subr{'-'};
1240 This finishes implementation of a primitive symbolic calculator in
1241 50 lines of Perl code. Since the numeric values of subexpressions
1242 are not cached, the calculator is very slow.
1244 Here is the answer for the exercise: In the case of str(), we need no
1245 explicit recursion since the overloaded C<.>-operator will fall back
1246 to an existing overloaded operator C<"">. Overloaded arithmetic
1247 operators I<do not> fall back to numeric conversion if C<fallback> is
1248 not explicitly requested. Thus without an explicit recursion num()
1249 would convert C<['+', $a, $b]> to C<$a + $b>, which would just rebuild
1250 the argument of num().
1252 If you wonder why defaults for conversion are different for str() and
1253 num(), note how easy it was to write the symbolic calculator. This
1254 simplicity is due to an appropriate choice of defaults. One extra
1255 note: due to the explicit recursion num() is more fragile than sym():
1256 we need to explicitly check for the type of $a and $b. If components
1257 $a and $b happen to be of some related type, this may lead to problems.
1259 =head2 I<Really> symbolic calculator
1261 One may wonder why we call the above calculator symbolic. The reason
1262 is that the actual calculation of the value of expression is postponed
1263 until the value is I<used>.
1265 To see it in action, add a method
1270 @$obj->[0,1] = ('=', shift);
1273 to the package C<symbolic>. After this change one can do
1275 my $a = new symbolic 3;
1276 my $b = new symbolic 4;
1277 my $c = sqrt($a**2 + $b**2);
1279 and the numeric value of $c becomes 5. However, after calling
1281 $a->STORE(12); $b->STORE(5);
1283 the numeric value of $c becomes 13. There is no doubt now that the module
1284 symbolic provides a I<symbolic> calculator indeed.
1286 To hide the rough edges under the hood, provide a tie()d interface to the
1287 package C<symbolic> (compare with L<Metaphor clash>). Add methods
1289 sub TIESCALAR { my $pack = shift; $pack->new(@_) }
1291 sub nop { } # Around a bug
1293 (the bug is described in L<"BUGS">). One can use this new interface as
1295 tie $a, 'symbolic', 3;
1296 tie $b, 'symbolic', 4;
1297 $a->nop; $b->nop; # Around a bug
1299 my $c = sqrt($a**2 + $b**2);
1301 Now numeric value of $c is 5. After C<$a = 12; $b = 5> the numeric value
1302 of $c becomes 13. To insulate the user of the module add a method
1304 sub vars { my $p = shift; tie($_, $p), $_->nop foreach @_; }
1309 symbolic->vars($a, $b);
1310 my $c = sqrt($a**2 + $b**2);
1313 printf "c5 %s=%f\n", $c, $c;
1316 printf "c13 %s=%f\n", $c, $c;
1318 shows that the numeric value of $c follows changes to the values of $a
1323 Ilya Zakharevich E<lt>F<ilya@math.mps.ohio-state.edu>E<gt>.
1327 When Perl is run with the B<-Do> switch or its equivalent, overloading
1328 induces diagnostic messages.
1330 Using the C<m> command of Perl debugger (see L<perldebug>) one can
1331 deduce which operations are overloaded (and which ancestor triggers
1332 this overloading). Say, if C<eq> is overloaded, then the method C<(eq>
1333 is shown by debugger. The method C<()> corresponds to the C<fallback>
1334 key (in fact a presence of this method shows that this package has
1335 overloading enabled, and it is what is used by the C<Overloaded>
1336 function of module C<overload>).
1340 Because it is used for overloading, the per-package hash %OVERLOAD now
1341 has a special meaning in Perl. The symbol table is filled with names
1342 looking like line-noise.
1344 For the purpose of inheritance every overloaded package behaves as if
1345 C<fallback> is present (possibly undefined). This may create
1346 interesting effects if some package is not overloaded, but inherits
1347 from two overloaded packages.
1349 Relation between overloading and tie()ing is broken. Overloading is
1350 triggered or not basing on the I<previous> class of tie()d value.
1352 This happens because the presence of overloading is checked too early,
1353 before any tie()d access is attempted. If the FETCH()ed class of the
1354 tie()d value does not change, a simple workaround is to access the value
1355 immediately after tie()ing, so that after this call the I<previous> class
1356 coincides with the current one.
1358 B<Needed:> a way to fix this without a speed penalty.
1360 Barewords are not covered by overloaded string constants.
1362 This document is confusing. There are grammos and misleading language
1363 used in places. It would seem a total rewrite is needed.