3 $overload::hint_bits = 0x20000;
11 $ {$package . "::OVERLOAD"}{dummy}++; # Register with magic by touching.
12 *{$package . "::()"} = \&nil; # Make it findable via fetchmethod.
14 if ($_ eq 'fallback') {
18 if (not ref $sub and $sub !~ /::/) {
19 $ {$package . "::(" . $_} = $sub;
22 #print STDERR "Setting `$ {'package'}::\cO$_' to \\&`$sub'.\n";
23 *{$package . "::(" . $_} = \&{ $sub };
26 ${$package . "::()"} = $fb; # Make it findable too (fallback only).
30 $package = (caller())[0];
31 # *{$package . "::OVERLOAD"} = \&OVERLOAD;
33 $package->overload::OVERLOAD(@_);
37 $package = (caller())[0];
38 ${$package . "::OVERLOAD"}{dummy}++; # Upgrade the table
41 if ($_ eq 'fallback') {
42 undef $ {$package . "::()"};
44 delete $ {$package . "::"}{"(" . $_};
51 $package = ref $package if ref $package;
57 return undef unless $globref;
58 my $sub = \&{*$globref};
59 return $sub if $sub ne \&nil;
60 return shift->can($ {*$globref});
63 sub OverloadedStringify {
65 $package = ref $package if ref $package;
67 ov_method mycan($package, '(""'), $package
68 or ov_method mycan($package, '(0+'), $package
69 or ov_method mycan($package, '(bool'), $package
70 or ov_method mycan($package, '(nomethod'), $package;
75 $package = ref $package if ref $package;
76 #my $meth = $package->can('(' . shift);
77 ov_method mycan($package, '(' . shift), $package;
78 #return $meth if $meth ne \&nil;
83 my $package = ref $_[0];
84 return "$_[0]" unless $package;
85 bless $_[0], overload::Fake; # Non-overloaded package
87 bless $_[0], $package; # Back
88 $package . substr $str, index $str, '=';
92 (OverloadedStringify($_[0]) or ref($_[0]) eq 'Regexp') ?
97 sub mycan { # Real can would leave stubs.
98 my ($package, $meth) = @_;
99 return \*{$package . "::$meth"} if defined &{$package . "::$meth"};
101 foreach $p (@{$package . "::ISA"}) {
102 my $out = mycan($p, $meth);
116 %ops = ( with_assign => "+ - * / % ** << >> x .",
117 assign => "+= -= *= /= %= **= <<= >>= x= .=",
118 num_comparison => "< <= > >= == !=",
119 '3way_comparison'=> "<=> cmp",
120 str_comparison => "lt le gt ge eq ne",
124 func => "atan2 cos sin exp abs log sqrt",
125 conversion => 'bool "" 0+',
127 dereferencing => '${} @{} %{} &{} *{}',
128 special => 'nomethod fallback =');
130 use warnings::register;
132 # Arguments: what, sub
135 if (warnings::enabled) {
137 Carp::carp ("Odd number of arguments for overload::constant");
141 elsif (!exists $constants {$_ [0]}) {
142 if (warnings::enabled) {
144 Carp::carp ("`$_[0]' is not an overloadable type");
147 elsif (!ref $_ [1] || "$_[1]" !~ /CODE\(0x[\da-f]+\)$/) {
148 # Can't use C<ref $_[1] eq "CODE"> above as code references can be
149 # blessed, and C<ref> would return the package the ref is blessed into.
150 if (warnings::enabled) {
152 $_ [1] = "undef" unless defined $_ [1];
153 Carp::carp ("`$_[1]' is not a code reference");
158 $^H |= $constants{$_[0]} | $overload::hint_bits;
164 sub remove_constant {
165 # Arguments: what, sub
168 $^H &= ~ $constants{$_[0]};
179 overload - Package for overloading perl operations
192 $a = new SomeThing 57;
195 if (overload::Overloaded $b) {...}
197 $strval = overload::StrVal $b;
201 =head2 Declaration of overloaded functions
203 The compilation directive
210 declares function Number::add() for addition, and method muas() in
211 the "class" C<Number> (or one of its base classes)
212 for the assignment form C<*=> of multiplication.
214 Arguments of this directive come in (key, value) pairs. Legal values
215 are values legal inside a C<&{ ... }> call, so the name of a
216 subroutine, a reference to a subroutine, or an anonymous subroutine
217 will all work. Note that values specified as strings are
218 interpreted as methods, not subroutines. Legal keys are listed below.
220 The subroutine C<add> will be called to execute C<$a+$b> if $a
221 is a reference to an object blessed into the package C<Number>, or if $a is
222 not an object from a package with defined mathemagic addition, but $b is a
223 reference to a C<Number>. It can also be called in other situations, like
224 C<$a+=7>, or C<$a++>. See L<MAGIC AUTOGENERATION>. (Mathemagical
225 methods refer to methods triggered by an overloaded mathematical
228 Since overloading respects inheritance via the @ISA hierarchy, the
229 above declaration would also trigger overloading of C<+> and C<*=> in
230 all the packages which inherit from C<Number>.
232 =head2 Calling Conventions for Binary Operations
234 The functions specified in the C<use overload ...> directive are called
235 with three (in one particular case with four, see L<Last Resort>)
236 arguments. If the corresponding operation is binary, then the first
237 two arguments are the two arguments of the operation. However, due to
238 general object calling conventions, the first argument should always be
239 an object in the package, so in the situation of C<7+$a>, the
240 order of the arguments is interchanged. It probably does not matter
241 when implementing the addition method, but whether the arguments
242 are reversed is vital to the subtraction method. The method can
243 query this information by examining the third argument, which can take
244 three different values:
250 the order of arguments is as in the current operation.
254 the arguments are reversed.
258 the current operation is an assignment variant (as in
259 C<$a+=7>), but the usual function is called instead. This additional
260 information can be used to generate some optimizations. Compare
261 L<Calling Conventions for Mutators>.
265 =head2 Calling Conventions for Unary Operations
267 Unary operation are considered binary operations with the second
268 argument being C<undef>. Thus the functions that overloads C<{"++"}>
269 is called with arguments C<($a,undef,'')> when $a++ is executed.
271 =head2 Calling Conventions for Mutators
273 Two types of mutators have different calling conventions:
277 =item C<++> and C<-->
279 The routines which implement these operators are expected to actually
280 I<mutate> their arguments. So, assuming that $obj is a reference to a
283 sub incr { my $n = $ {$_[0]}; ++$n; $_[0] = bless \$n}
285 is an appropriate implementation of overloaded C<++>. Note that
287 sub incr { ++$ {$_[0]} ; shift }
289 is OK if used with preincrement and with postincrement. (In the case
290 of postincrement a copying will be performed, see L<Copy Constructor>.)
292 =item C<x=> and other assignment versions
294 There is nothing special about these methods. They may change the
295 value of their arguments, and may leave it as is. The result is going
296 to be assigned to the value in the left-hand-side if different from
299 This allows for the same method to be used as overloaded C<+=> and
300 C<+>. Note that this is I<allowed>, but not recommended, since by the
301 semantic of L<"Fallback"> Perl will call the method for C<+> anyway,
302 if C<+=> is not overloaded.
306 B<Warning.> Due to the presense of assignment versions of operations,
307 routines which may be called in assignment context may create
308 self-referential structures. Currently Perl will not free self-referential
309 structures until cycles are C<explicitly> broken. You may get problems
310 when traversing your structures too.
314 use overload '+' => sub { bless [ \$_[0], \$_[1] ] };
316 is asking for trouble, since for code C<$obj += $foo> the subroutine
317 is called as C<$obj = add($obj, $foo, undef)>, or C<$obj = [\$obj,
318 \$foo]>. If using such a subroutine is an important optimization, one
319 can overload C<+=> explicitly by a non-"optimized" version, or switch
320 to non-optimized version if C<not defined $_[2]> (see
321 L<Calling Conventions for Binary Operations>).
323 Even if no I<explicit> assignment-variants of operators are present in
324 the script, they may be generated by the optimizer. Say, C<",$obj,"> or
325 C<',' . $obj . ','> may be both optimized to
327 my $tmp = ',' . $obj; $tmp .= ',';
329 =head2 Overloadable Operations
331 The following symbols can be specified in C<use overload> directive:
335 =item * I<Arithmetic operations>
337 "+", "+=", "-", "-=", "*", "*=", "/", "/=", "%", "%=",
338 "**", "**=", "<<", "<<=", ">>", ">>=", "x", "x=", ".", ".=",
340 For these operations a substituted non-assignment variant can be called if
341 the assignment variant is not available. Methods for operations "C<+>",
342 "C<->", "C<+=>", and "C<-=>" can be called to automatically generate
343 increment and decrement methods. The operation "C<->" can be used to
344 autogenerate missing methods for unary minus or C<abs>.
346 See L<"MAGIC AUTOGENERATION">, L<"Calling Conventions for Mutators"> and
347 L<"Calling Conventions for Binary Operations">) for details of these
350 =item * I<Comparison operations>
352 "<", "<=", ">", ">=", "==", "!=", "<=>",
353 "lt", "le", "gt", "ge", "eq", "ne", "cmp",
355 If the corresponding "spaceship" variant is available, it can be
356 used to substitute for the missing operation. During C<sort>ing
357 arrays, C<cmp> is used to compare values subject to C<use overload>.
359 =item * I<Bit operations>
361 "&", "^", "|", "neg", "!", "~",
363 "C<neg>" stands for unary minus. If the method for C<neg> is not
364 specified, it can be autogenerated using the method for
365 subtraction. If the method for "C<!>" is not specified, it can be
366 autogenerated using the methods for "C<bool>", or "C<\"\">", or "C<0+>".
368 =item * I<Increment and decrement>
372 If undefined, addition and subtraction methods can be
373 used instead. These operations are called both in prefix and
376 =item * I<Transcendental functions>
378 "atan2", "cos", "sin", "exp", "abs", "log", "sqrt",
380 If C<abs> is unavailable, it can be autogenerated using methods
381 for "E<lt>" or "E<lt>=E<gt>" combined with either unary minus or subtraction.
383 =item * I<Boolean, string and numeric conversion>
385 "bool", "\"\"", "0+",
387 If one or two of these operations are not overloaded, the remaining ones can
388 be used instead. C<bool> is used in the flow control operators
389 (like C<while>) and for the ternary "C<?:>" operation. These functions can
390 return any arbitrary Perl value. If the corresponding operation for this value
391 is overloaded too, that operation will be called again with this value.
397 If not overloaded, the argument will be converted to a filehandle or
398 glob (which may require a stringification). The same overloading
399 happens both for the I<read-filehandle> syntax C<E<lt>$varE<gt>> and
400 I<globbing> syntax C<E<lt>${var}E<gt>>.
402 =item * I<Dereferencing>
404 '${}', '@{}', '%{}', '&{}', '*{}'.
406 If not overloaded, the argument will be dereferenced I<as is>, thus
407 should be of correct type. These functions should return a reference
408 of correct type, or another object with overloaded dereferencing.
412 "nomethod", "fallback", "=",
414 see L<SPECIAL SYMBOLS FOR C<use overload>>.
418 See L<"Fallback"> for an explanation of when a missing method can be
421 A computer-readable form of the above table is available in the hash
422 %overload::ops, with values being space-separated lists of names:
424 with_assign => '+ - * / % ** << >> x .',
425 assign => '+= -= *= /= %= **= <<= >>= x= .=',
426 num_comparison => '< <= > >= == !=',
427 '3way_comparison'=> '<=> cmp',
428 str_comparison => 'lt le gt ge eq ne',
432 func => 'atan2 cos sin exp abs log sqrt',
433 conversion => 'bool "" 0+',
435 dereferencing => '${} @{} %{} &{} *{}',
436 special => 'nomethod fallback ='
438 =head2 Inheritance and overloading
440 Inheritance interacts with overloading in two ways.
444 =item Strings as values of C<use overload> directive
448 use overload key => value;
450 is a string, it is interpreted as a method name.
452 =item Overloading of an operation is inherited by derived classes
454 Any class derived from an overloaded class is also overloaded. The
455 set of overloaded methods is the union of overloaded methods of all
456 the ancestors. If some method is overloaded in several ancestor, then
457 which description will be used is decided by the usual inheritance
460 If C<A> inherits from C<B> and C<C> (in this order), C<B> overloads
461 C<+> with C<\&D::plus_sub>, and C<C> overloads C<+> by C<"plus_meth">,
462 then the subroutine C<D::plus_sub> will be called to implement
463 operation C<+> for an object in package C<A>.
467 Note that since the value of the C<fallback> key is not a subroutine,
468 its inheritance is not governed by the above rules. In the current
469 implementation, the value of C<fallback> in the first overloaded
470 ancestor is used, but this is accidental and subject to change.
472 =head1 SPECIAL SYMBOLS FOR C<use overload>
474 Three keys are recognized by Perl that are not covered by the above
479 C<"nomethod"> should be followed by a reference to a function of four
480 parameters. If defined, it is called when the overloading mechanism
481 cannot find a method for some operation. The first three arguments of
482 this function coincide with the arguments for the corresponding method if
483 it were found, the fourth argument is the symbol
484 corresponding to the missing method. If several methods are tried,
485 the last one is used. Say, C<1-$a> can be equivalent to
487 &nomethodMethod($a,1,1,"-")
489 if the pair C<"nomethod" =E<gt> "nomethodMethod"> was specified in the
490 C<use overload> directive.
492 If some operation cannot be resolved, and there is no function
493 assigned to C<"nomethod">, then an exception will be raised via die()--
494 unless C<"fallback"> was specified as a key in C<use overload> directive.
498 The key C<"fallback"> governs what to do if a method for a particular
499 operation is not found. Three different cases are possible depending on
500 the value of C<"fallback">:
507 substituted method (see L<MAGIC AUTOGENERATION>). If this fails, it
508 then tries to calls C<"nomethod"> value; if missing, an exception
513 The same as for the C<undef> value, but no exception is raised. Instead,
514 it silently reverts to what it would have done were there no C<use overload>
517 =item * defined, but FALSE
519 No autogeneration is tried. Perl tries to call
520 C<"nomethod"> value, and if this is missing, raises an exception.
524 B<Note.> C<"fallback"> inheritance via @ISA is not carved in stone
525 yet, see L<"Inheritance and overloading">.
527 =head2 Copy Constructor
529 The value for C<"="> is a reference to a function with three
530 arguments, i.e., it looks like the other values in C<use
531 overload>. However, it does not overload the Perl assignment
532 operator. This would go against Camel hair.
534 This operation is called in the situations when a mutator is applied
535 to a reference that shares its object with some other reference, such
541 To make this change $a and not change $b, a copy of C<$$a> is made,
542 and $a is assigned a reference to this new object. This operation is
543 done during execution of the C<++$a>, and not during the assignment,
544 (so before the increment C<$$a> coincides with C<$$b>). This is only
545 done if C<++> is expressed via a method for C<'++'> or C<'+='> (or
546 C<nomethod>). Note that if this operation is expressed via C<'+'>
547 a nonmutator, i.e., as in
552 then C<$a> does not reference a new copy of C<$$a>, since $$a does not
553 appear as lvalue when the above code is executed.
555 If the copy constructor is required during the execution of some mutator,
556 but a method for C<'='> was not specified, it can be autogenerated as a
557 string copy if the object is a plain scalar.
563 The actually executed code for
566 Something else which does not modify $a or $b....
572 Something else which does not modify $a or $b....
573 $a = $a->clone(undef,"");
576 if $b was mathemagical, and C<'++'> was overloaded with C<\&incr>,
577 C<'='> was overloaded with C<\&clone>.
581 Same behaviour is triggered by C<$b = $a++>, which is consider a synonym for
584 =head1 MAGIC AUTOGENERATION
586 If a method for an operation is not found, and the value for C<"fallback"> is
587 TRUE or undefined, Perl tries to autogenerate a substitute method for
588 the missing operation based on the defined operations. Autogenerated method
589 substitutions are possible for the following operations:
593 =item I<Assignment forms of arithmetic operations>
595 C<$a+=$b> can use the method for C<"+"> if the method for C<"+=">
598 =item I<Conversion operations>
600 String, numeric, and boolean conversion are calculated in terms of one
601 another if not all of them are defined.
603 =item I<Increment and decrement>
605 The C<++$a> operation can be expressed in terms of C<$a+=1> or C<$a+1>,
606 and C<$a--> in terms of C<$a-=1> and C<$a-1>.
610 can be expressed in terms of C<$aE<lt>0> and C<-$a> (or C<0-$a>).
614 can be expressed in terms of subtraction.
618 C<!> and C<not> can be expressed in terms of boolean conversion, or
619 string or numerical conversion.
621 =item I<Concatenation>
623 can be expressed in terms of string conversion.
625 =item I<Comparison operations>
627 can be expressed in terms of its "spaceship" counterpart: either
628 C<E<lt>=E<gt>> or C<cmp>:
630 <, >, <=, >=, ==, != in terms of <=>
631 lt, gt, le, ge, eq, ne in terms of cmp
635 <> in terms of builtin operations
637 =item I<Dereferencing>
639 ${} @{} %{} &{} *{} in terms of builtin operations
641 =item I<Copy operator>
643 can be expressed in terms of an assignment to the dereferenced value, if this
644 value is a scalar and not a reference.
648 =head1 Losing overloading
650 The restriction for the comparison operation is that even if, for example,
651 `C<cmp>' should return a blessed reference, the autogenerated `C<lt>'
652 function will produce only a standard logical value based on the
653 numerical value of the result of `C<cmp>'. In particular, a working
654 numeric conversion is needed in this case (possibly expressed in terms of
657 Similarly, C<.=> and C<x=> operators lose their mathemagical properties
658 if the string conversion substitution is applied.
660 When you chop() a mathemagical object it is promoted to a string and its
661 mathemagical properties are lost. The same can happen with other
664 =head1 Run-time Overloading
666 Since all C<use> directives are executed at compile-time, the only way to
667 change overloading during run-time is to
669 eval 'use overload "+" => \&addmethod';
673 eval 'no overload "+", "--", "<="';
675 though the use of these constructs during run-time is questionable.
677 =head1 Public functions
679 Package C<overload.pm> provides the following public functions:
683 =item overload::StrVal(arg)
685 Gives string value of C<arg> as in absence of stringify overloading.
687 =item overload::Overloaded(arg)
689 Returns true if C<arg> is subject to overloading of some operations.
691 =item overload::Method(obj,op)
693 Returns C<undef> or a reference to the method that implements C<op>.
697 =head1 Overloading constants
699 For some application Perl parser mangles constants too much. It is possible
700 to hook into this process via overload::constant() and overload::remove_constant()
703 These functions take a hash as an argument. The recognized keys of this hash
710 to overload integer constants,
714 to overload floating point constants,
718 to overload octal and hexadecimal constants,
722 to overload C<q>-quoted strings, constant pieces of C<qq>- and C<qx>-quoted
723 strings and here-documents,
727 to overload constant pieces of regular expressions.
731 The corresponding values are references to functions which take three arguments:
732 the first one is the I<initial> string form of the constant, the second one
733 is how Perl interprets this constant, the third one is how the constant is used.
734 Note that the initial string form does not
735 contain string delimiters, and has backslashes in backslash-delimiter
736 combinations stripped (thus the value of delimiter is not relevant for
737 processing of this string). The return value of this function is how this
738 constant is going to be interpreted by Perl. The third argument is undefined
739 unless for overloaded C<q>- and C<qr>- constants, it is C<q> in single-quote
740 context (comes from strings, regular expressions, and single-quote HERE
741 documents), it is C<tr> for arguments of C<tr>/C<y> operators,
742 it is C<s> for right-hand side of C<s>-operator, and it is C<qq> otherwise.
744 Since an expression C<"ab$cd,,"> is just a shortcut for C<'ab' . $cd . ',,'>,
745 it is expected that overloaded constant strings are equipped with reasonable
746 overloaded catenation operator, otherwise absurd results will result.
747 Similarly, negative numbers are considered as negations of positive constants.
749 Note that it is probably meaningless to call the functions overload::constant()
750 and overload::remove_constant() from anywhere but import() and unimport() methods.
751 From these methods they may be called as
756 die "unknown import: @_" unless @_ == 1 and $_[0] eq ':constant';
757 overload::constant integer => sub {Math::BigInt->new(shift)};
760 B<BUGS> Currently overloaded-ness of constants does not propagate
763 =head1 IMPLEMENTATION
765 What follows is subject to change RSN.
767 The table of methods for all operations is cached in magic for the
768 symbol table hash for the package. The cache is invalidated during
769 processing of C<use overload>, C<no overload>, new function
770 definitions, and changes in @ISA. However, this invalidation remains
771 unprocessed until the next C<bless>ing into the package. Hence if you
772 want to change overloading structure dynamically, you'll need an
773 additional (fake) C<bless>ing to update the table.
775 (Every SVish thing has a magic queue, and magic is an entry in that
776 queue. This is how a single variable may participate in multiple
777 forms of magic simultaneously. For instance, environment variables
778 regularly have two forms at once: their %ENV magic and their taint
779 magic. However, the magic which implements overloading is applied to
780 the stashes, which are rarely used directly, thus should not slow down
783 If an object belongs to a package using overload, it carries a special
784 flag. Thus the only speed penalty during arithmetic operations without
785 overloading is the checking of this flag.
787 In fact, if C<use overload> is not present, there is almost no overhead
788 for overloadable operations, so most programs should not suffer
789 measurable performance penalties. A considerable effort was made to
790 minimize the overhead when overload is used in some package, but the
791 arguments in question do not belong to packages using overload. When
792 in doubt, test your speed with C<use overload> and without it. So far
793 there have been no reports of substantial speed degradation if Perl is
794 compiled with optimization turned on.
796 There is no size penalty for data if overload is not used. The only
797 size penalty if overload is used in some package is that I<all> the
798 packages acquire a magic during the next C<bless>ing into the
799 package. This magic is three-words-long for packages without
800 overloading, and carries the cache table if the package is overloaded.
802 Copying (C<$a=$b>) is shallow; however, a one-level-deep copying is
803 carried out before any operation that can imply an assignment to the
804 object $a (or $b) refers to, like C<$a++>. You can override this
805 behavior by defining your own copy constructor (see L<"Copy Constructor">).
807 It is expected that arguments to methods that are not explicitly supposed
808 to be changed are constant (but this is not enforced).
810 =head1 Metaphor clash
812 One may wonder why the semantic of overloaded C<=> is so counter intuitive.
813 If it I<looks> counter intuitive to you, you are subject to a metaphor
816 Here is a Perl object metaphor:
818 I< object is a reference to blessed data>
820 and an arithmetic metaphor:
822 I< object is a thing by itself>.
824 The I<main> problem of overloading C<=> is the fact that these metaphors
825 imply different actions on the assignment C<$a = $b> if $a and $b are
826 objects. Perl-think implies that $a becomes a reference to whatever
827 $b was referencing. Arithmetic-think implies that the value of "object"
828 $a is changed to become the value of the object $b, preserving the fact
829 that $a and $b are separate entities.
831 The difference is not relevant in the absence of mutators. After
832 a Perl-way assignment an operation which mutates the data referenced by $a
833 would change the data referenced by $b too. Effectively, after
834 C<$a = $b> values of $a and $b become I<indistinguishable>.
836 On the other hand, anyone who has used algebraic notation knows the
837 expressive power of the arithmetic metaphor. Overloading works hard
838 to enable this metaphor while preserving the Perlian way as far as
839 possible. Since it is not not possible to freely mix two contradicting
840 metaphors, overloading allows the arithmetic way to write things I<as
841 far as all the mutators are called via overloaded access only>. The
842 way it is done is described in L<Copy Constructor>.
844 If some mutator methods are directly applied to the overloaded values,
845 one may need to I<explicitly unlink> other values which references the
850 $b = $a; # $b is "linked" to $a
852 $a = $a->clone; # Unlink $b from $a
855 Note that overloaded access makes this transparent:
858 $b = $a; # $b is "linked" to $a
859 $a += 4; # would unlink $b automagically
861 However, it would not make
864 $a = 4; # Now $a is a plain 4, not 'Data'
866 preserve "objectness" of $a. But Perl I<has> a way to make assignments
867 to an object do whatever you want. It is just not the overload, but
868 tie()ing interface (see L<perlfunc/tie>). Adding a FETCH() method
869 which returns the object itself, and STORE() method which changes the
870 value of the object, one can reproduce the arithmetic metaphor in its
871 completeness, at least for variables which were tie()d from the start.
873 (Note that a workaround for a bug may be needed, see L<"BUGS">.)
877 Please add examples to what follows!
879 =head2 Two-face scalars
881 Put this in F<two_face.pm> in your Perl library directory:
883 package two_face; # Scalars with separate string and
885 sub new { my $p = shift; bless [@_], $p }
886 use overload '""' => \&str, '0+' => \&num, fallback => 1;
893 my $seven = new two_face ("vii", 7);
894 printf "seven=$seven, seven=%d, eight=%d\n", $seven, $seven+1;
895 print "seven contains `i'\n" if $seven =~ /i/;
897 (The second line creates a scalar which has both a string value, and a
898 numeric value.) This prints:
900 seven=vii, seven=7, eight=8
903 =head2 Two-face references
905 Suppose you want to create an object which is accessible as both an
906 array reference and a hash reference, similar to the
907 L<pseudo-hash|perlref/"Pseudo-hashes: Using an array as a hash">
908 builtin Perl type. Let's make it better than a pseudo-hash by
909 allowing index 0 to be treated as a normal element.
912 use overload '%{}' => \&gethash, '@{}' => sub { $ {shift()} };
920 tie %h, ref $self, $self;
924 sub TIEHASH { my $p = shift; bless \ shift, $p }
927 $fields{$_} = $i++ foreach qw{zero one two three};
929 my $self = ${shift()};
930 my $key = $fields{shift()};
931 defined $key or die "Out of band access";
932 $$self->[$key] = shift;
935 my $self = ${shift()};
936 my $key = $fields{shift()};
937 defined $key or die "Out of band access";
941 Now one can access an object using both the array and hash syntax:
943 my $bar = new two_refs 3,4,5,6;
945 $bar->{two} == 11 or die 'bad hash fetch';
947 Note several important features of this example. First of all, the
948 I<actual> type of $bar is a scalar reference, and we do not overload
949 the scalar dereference. Thus we can get the I<actual> non-overloaded
950 contents of $bar by just using C<$$bar> (what we do in functions which
951 overload dereference). Similarly, the object returned by the
952 TIEHASH() method is a scalar reference.
954 Second, we create a new tied hash each time the hash syntax is used.
955 This allows us not to worry about a possibility of a reference loop,
956 would would lead to a memory leak.
958 Both these problems can be cured. Say, if we want to overload hash
959 dereference on a reference to an object which is I<implemented> as a
960 hash itself, the only problem one has to circumvent is how to access
961 this I<actual> hash (as opposed to the I<virtual> exhibited by
962 overloaded dereference operator). Here is one possible fetching routine:
965 my ($self, $key) = (shift, shift);
966 my $class = ref $self;
967 bless $self, 'overload::dummy'; # Disable overloading of %{}
968 my $out = $self->{$key};
969 bless $self, $class; # Restore overloading
973 To move creation of the tied hash on each access, one may an extra
974 level of indirection which allows a non-circular structure of references:
977 use overload '%{}' => sub { ${shift()}->[1] },
978 '@{}' => sub { ${shift()}->[0] };
984 bless \ [$a, \%h], $p;
989 tie %h, ref $self, $self;
993 sub TIEHASH { my $p = shift; bless \ shift, $p }
996 $fields{$_} = $i++ foreach qw{zero one two three};
999 my $key = $fields{shift()};
1000 defined $key or die "Out of band access";
1005 my $key = $fields{shift()};
1006 defined $key or die "Out of band access";
1010 Now if $baz is overloaded like this, then C<$bar> is a reference to a
1011 reference to the intermediate array, which keeps a reference to an
1012 actual array, and the access hash. The tie()ing object for the access
1013 hash is also a reference to a reference to the actual array, so
1019 There are no loops of references.
1023 Both "objects" which are blessed into the class C<two_refs1> are
1024 references to a reference to an array, thus references to a I<scalar>.
1025 Thus the accessor expression C<$$foo-E<gt>[$ind]> involves no
1026 overloaded operations.
1030 =head2 Symbolic calculator
1032 Put this in F<symbolic.pm> in your Perl library directory:
1034 package symbolic; # Primitive symbolic calculator
1035 use overload nomethod => \&wrap;
1037 sub new { shift; bless ['n', @_] }
1039 my ($obj, $other, $inv, $meth) = @_;
1040 ($obj, $other) = ($other, $obj) if $inv;
1041 bless [$meth, $obj, $other];
1044 This module is very unusual as overloaded modules go: it does not
1045 provide any usual overloaded operators, instead it provides the L<Last
1046 Resort> operator C<nomethod>. In this example the corresponding
1047 subroutine returns an object which encapsulates operations done over
1048 the objects: C<new symbolic 3> contains C<['n', 3]>, C<2 + new
1049 symbolic 3> contains C<['+', 2, ['n', 3]]>.
1051 Here is an example of the script which "calculates" the side of
1052 circumscribed octagon using the above package:
1055 my $iter = 1; # 2**($iter+2) = 8
1056 my $side = new symbolic 1;
1060 $side = (sqrt(1 + $side**2) - 1)/$side;
1064 The value of $side is
1066 ['/', ['-', ['sqrt', ['+', 1, ['**', ['n', 1], 2]],
1067 undef], 1], ['n', 1]]
1069 Note that while we obtained this value using a nice little script,
1070 there is no simple way to I<use> this value. In fact this value may
1071 be inspected in debugger (see L<perldebug>), but ony if
1072 C<bareStringify> B<O>ption is set, and not via C<p> command.
1074 If one attempts to print this value, then the overloaded operator
1075 C<""> will be called, which will call C<nomethod> operator. The
1076 result of this operator will be stringified again, but this result is
1077 again of type C<symbolic>, which will lead to an infinite loop.
1079 Add a pretty-printer method to the module F<symbolic.pm>:
1082 my ($meth, $a, $b) = @{+shift};
1083 $a = 'u' unless defined $a;
1084 $b = 'u' unless defined $b;
1085 $a = $a->pretty if ref $a;
1086 $b = $b->pretty if ref $b;
1090 Now one can finish the script by
1092 print "side = ", $side->pretty, "\n";
1094 The method C<pretty> is doing object-to-string conversion, so it
1095 is natural to overload the operator C<""> using this method. However,
1096 inside such a method it is not necessary to pretty-print the
1097 I<components> $a and $b of an object. In the above subroutine
1098 C<"[$meth $a $b]"> is a catenation of some strings and components $a
1099 and $b. If these components use overloading, the catenation operator
1100 will look for an overloaded operator C<.>, if not present, it will
1101 look for an overloaded operator C<"">. Thus it is enough to use
1103 use overload nomethod => \&wrap, '""' => \&str;
1105 my ($meth, $a, $b) = @{+shift};
1106 $a = 'u' unless defined $a;
1107 $b = 'u' unless defined $b;
1111 Now one can change the last line of the script to
1113 print "side = $side\n";
1117 side = [/ [- [sqrt [+ 1 [** [n 1 u] 2]] u] 1] [n 1 u]]
1119 and one can inspect the value in debugger using all the possible
1122 Something is is still amiss: consider the loop variable $cnt of the
1123 script. It was a number, not an object. We cannot make this value of
1124 type C<symbolic>, since then the loop will not terminate.
1126 Indeed, to terminate the cycle, the $cnt should become false.
1127 However, the operator C<bool> for checking falsity is overloaded (this
1128 time via overloaded C<"">), and returns a long string, thus any object
1129 of type C<symbolic> is true. To overcome this, we need a way to
1130 compare an object to 0. In fact, it is easier to write a numeric
1133 Here is the text of F<symbolic.pm> with such a routine added (and
1134 slightly modified str()):
1136 package symbolic; # Primitive symbolic calculator
1138 nomethod => \&wrap, '""' => \&str, '0+' => \#
1140 sub new { shift; bless ['n', @_] }
1142 my ($obj, $other, $inv, $meth) = @_;
1143 ($obj, $other) = ($other, $obj) if $inv;
1144 bless [$meth, $obj, $other];
1147 my ($meth, $a, $b) = @{+shift};
1148 $a = 'u' unless defined $a;
1155 my %subr = ( n => sub {$_[0]},
1156 sqrt => sub {sqrt $_[0]},
1157 '-' => sub {shift() - shift()},
1158 '+' => sub {shift() + shift()},
1159 '/' => sub {shift() / shift()},
1160 '*' => sub {shift() * shift()},
1161 '**' => sub {shift() ** shift()},
1164 my ($meth, $a, $b) = @{+shift};
1165 my $subr = $subr{$meth}
1166 or die "Do not know how to ($meth) in symbolic";
1167 $a = $a->num if ref $a eq __PACKAGE__;
1168 $b = $b->num if ref $b eq __PACKAGE__;
1172 All the work of numeric conversion is done in %subr and num(). Of
1173 course, %subr is not complete, it contains only operators used in the
1174 example below. Here is the extra-credit question: why do we need an
1175 explicit recursion in num()? (Answer is at the end of this section.)
1177 Use this module like this:
1180 my $iter = new symbolic 2; # 16-gon
1181 my $side = new symbolic 1;
1185 $cnt = $cnt - 1; # Mutator `--' not implemented
1186 $side = (sqrt(1 + $side**2) - 1)/$side;
1188 printf "%s=%f\n", $side, $side;
1189 printf "pi=%f\n", $side*(2**($iter+2));
1191 It prints (without so many line breaks)
1193 [/ [- [sqrt [+ 1 [** [/ [- [sqrt [+ 1 [** [n 1] 2]]] 1]
1195 [/ [- [sqrt [+ 1 [** [n 1] 2]]] 1] [n 1]]]=0.198912
1198 The above module is very primitive. It does not implement
1199 mutator methods (C<++>, C<-=> and so on), does not do deep copying
1200 (not required without mutators!), and implements only those arithmetic
1201 operations which are used in the example.
1203 To implement most arithmetic operations is easy, one should just use
1204 the tables of operations, and change the code which fills %subr to
1206 my %subr = ( 'n' => sub {$_[0]} );
1207 foreach my $op (split " ", $overload::ops{with_assign}) {
1208 $subr{$op} = $subr{"$op="} = eval "sub {shift() $op shift()}";
1210 my @bins = qw(binary 3way_comparison num_comparison str_comparison);
1211 foreach my $op (split " ", "@overload::ops{ @bins }") {
1212 $subr{$op} = eval "sub {shift() $op shift()}";
1214 foreach my $op (split " ", "@overload::ops{qw(unary func)}") {
1215 print "defining `$op'\n";
1216 $subr{$op} = eval "sub {$op shift()}";
1219 Due to L<Calling Conventions for Mutators>, we do not need anything
1220 special to make C<+=> and friends work, except filling C<+=> entry of
1221 %subr, and defining a copy constructor (needed since Perl has no
1222 way to know that the implementation of C<'+='> does not mutate
1223 the argument, compare L<Copy Constructor>).
1225 To implement a copy constructor, add C<'=' => \&cpy> to C<use overload>
1226 line, and code (this code assumes that mutators change things one level
1227 deep only, so recursive copying is not needed):
1231 bless [@$self], ref $self;
1234 To make C<++> and C<--> work, we need to implement actual mutators,
1235 either directly, or in C<nomethod>. We continue to do things inside
1236 C<nomethod>, thus add
1238 if ($meth eq '++' or $meth eq '--') {
1239 @$obj = ($meth, (bless [@$obj]), 1); # Avoid circular reference
1243 after the first line of wrap(). This is not a most effective
1244 implementation, one may consider
1246 sub inc { $_[0] = bless ['++', shift, 1]; }
1250 As a final remark, note that one can fill %subr by
1252 my %subr = ( 'n' => sub {$_[0]} );
1253 foreach my $op (split " ", $overload::ops{with_assign}) {
1254 $subr{$op} = $subr{"$op="} = eval "sub {shift() $op shift()}";
1256 my @bins = qw(binary 3way_comparison num_comparison str_comparison);
1257 foreach my $op (split " ", "@overload::ops{ @bins }") {
1258 $subr{$op} = eval "sub {shift() $op shift()}";
1260 foreach my $op (split " ", "@overload::ops{qw(unary func)}") {
1261 $subr{$op} = eval "sub {$op shift()}";
1263 $subr{'++'} = $subr{'+'};
1264 $subr{'--'} = $subr{'-'};
1266 This finishes implementation of a primitive symbolic calculator in
1267 50 lines of Perl code. Since the numeric values of subexpressions
1268 are not cached, the calculator is very slow.
1270 Here is the answer for the exercise: In the case of str(), we need no
1271 explicit recursion since the overloaded C<.>-operator will fall back
1272 to an existing overloaded operator C<"">. Overloaded arithmetic
1273 operators I<do not> fall back to numeric conversion if C<fallback> is
1274 not explicitly requested. Thus without an explicit recursion num()
1275 would convert C<['+', $a, $b]> to C<$a + $b>, which would just rebuild
1276 the argument of num().
1278 If you wonder why defaults for conversion are different for str() and
1279 num(), note how easy it was to write the symbolic calculator. This
1280 simplicity is due to an appropriate choice of defaults. One extra
1281 note: due to the explicit recursion num() is more fragile than sym():
1282 we need to explicitly check for the type of $a and $b. If components
1283 $a and $b happen to be of some related type, this may lead to problems.
1285 =head2 I<Really> symbolic calculator
1287 One may wonder why we call the above calculator symbolic. The reason
1288 is that the actual calculation of the value of expression is postponed
1289 until the value is I<used>.
1291 To see it in action, add a method
1296 @$obj->[0,1] = ('=', shift);
1299 to the package C<symbolic>. After this change one can do
1301 my $a = new symbolic 3;
1302 my $b = new symbolic 4;
1303 my $c = sqrt($a**2 + $b**2);
1305 and the numeric value of $c becomes 5. However, after calling
1307 $a->STORE(12); $b->STORE(5);
1309 the numeric value of $c becomes 13. There is no doubt now that the module
1310 symbolic provides a I<symbolic> calculator indeed.
1312 To hide the rough edges under the hood, provide a tie()d interface to the
1313 package C<symbolic> (compare with L<Metaphor clash>). Add methods
1315 sub TIESCALAR { my $pack = shift; $pack->new(@_) }
1317 sub nop { } # Around a bug
1319 (the bug is described in L<"BUGS">). One can use this new interface as
1321 tie $a, 'symbolic', 3;
1322 tie $b, 'symbolic', 4;
1323 $a->nop; $b->nop; # Around a bug
1325 my $c = sqrt($a**2 + $b**2);
1327 Now numeric value of $c is 5. After C<$a = 12; $b = 5> the numeric value
1328 of $c becomes 13. To insulate the user of the module add a method
1330 sub vars { my $p = shift; tie($_, $p), $_->nop foreach @_; }
1335 symbolic->vars($a, $b);
1336 my $c = sqrt($a**2 + $b**2);
1339 printf "c5 %s=%f\n", $c, $c;
1342 printf "c13 %s=%f\n", $c, $c;
1344 shows that the numeric value of $c follows changes to the values of $a
1349 Ilya Zakharevich E<lt>F<ilya@math.mps.ohio-state.edu>E<gt>.
1353 When Perl is run with the B<-Do> switch or its equivalent, overloading
1354 induces diagnostic messages.
1356 Using the C<m> command of Perl debugger (see L<perldebug>) one can
1357 deduce which operations are overloaded (and which ancestor triggers
1358 this overloading). Say, if C<eq> is overloaded, then the method C<(eq>
1359 is shown by debugger. The method C<()> corresponds to the C<fallback>
1360 key (in fact a presence of this method shows that this package has
1361 overloading enabled, and it is what is used by the C<Overloaded>
1362 function of module C<overload>).
1364 The module might issues the following warnings:
1368 =item Odd number of arguments for overload::constant
1370 (W) The call to overload::constant contained an odd number of arguments.
1371 The arguments should come in pairs.
1373 =item `%s' is not an overloadable type
1375 (W) You tried to overload a constant type the overload package is unaware of.
1377 =item `%s' is not a code reference
1379 (W) The second (fourth, sixth, ...) argument of overload::constant needs
1380 to be a code reference. Either an anonymous subroutine, or a reference
1387 Because it is used for overloading, the per-package hash %OVERLOAD now
1388 has a special meaning in Perl. The symbol table is filled with names
1389 looking like line-noise.
1391 For the purpose of inheritance every overloaded package behaves as if
1392 C<fallback> is present (possibly undefined). This may create
1393 interesting effects if some package is not overloaded, but inherits
1394 from two overloaded packages.
1396 Relation between overloading and tie()ing is broken. Overloading is
1397 triggered or not basing on the I<previous> class of tie()d value.
1399 This happens because the presence of overloading is checked too early,
1400 before any tie()d access is attempted. If the FETCH()ed class of the
1401 tie()d value does not change, a simple workaround is to access the value
1402 immediately after tie()ing, so that after this call the I<previous> class
1403 coincides with the current one.
1405 B<Needed:> a way to fix this without a speed penalty.
1407 Barewords are not covered by overloaded string constants.
1409 This document is confusing. There are grammos and misleading language
1410 used in places. It would seem a total rewrite is needed.