File Coverage

blib/lib/Class/C3.pm

Criterion	Covered	Total	%
statement	132	134	98.5
branch	33	38	86.8
condition	14	18	77.8
subroutine	29	30	96.7
pod	4	4	100.0
total	212	224	94.6

line	stmt	bran	cond	sub	pod	time	code
1
2							package Class::C3;
3
4	19			19		465	use strict;
	19					634
	19					483
5	19			19		332	use warnings;
	19					171
	19					475
6
7	19			19		300	use Scalar::Util 'blessed';
	19					168
	19					522
8	19			19		1063	use Algorithm::C3;
	19					197
	19					452
9
10							our $VERSION = '0.14';
11
12							# this is our global stash of both
13							# MRO's and method dispatch tables
14							# the structure basically looks like
15							# this:
16							#
17							# $MRO{$class} = {
18							# MRO => [ <class precendence list> ],
19							# methods => {
20							# orig => <original location of method>,
21							# code => \&<ref to original method>
22							# },
23							# has_overload_fallback => (1 \| 0)
24							# }
25							#
26							our %MRO;
27
28							# use these for debugging ...
29	0			0		0	sub _dump_MRO_table { %MRO }
30							our $TURN_OFF_C3 = 0;
31
32							# state tracking for initialize()/uninitialize()
33							our $_initialized = 0;
34
35							sub import {
36	96			96		1601	my $class = caller();
37							# skip if the caller is main::
38							# since that is clearly not relevant
39	96	100				1131	return if $class eq 'main';
40	79	50				783	return if $TURN_OFF_C3;
41							# make a note to calculate $class
42							# during INIT phase
43	79	50				1266	$MRO{$class} = undef unless exists $MRO{$class};
44							}
45
46							## initializers
47
48							sub initialize {
49							# why bother if we don't have anything ...
50	20	50		20	1	522	return unless keys %MRO;
51	20	100				280	if($_initialized) {
52	2					2709	uninitialize();
53	2					16	$MRO{$_} = undef foreach keys %MRO;
	2					56
54							}
55	20					338	_calculate_method_dispatch_tables();
56	20					288	_apply_method_dispatch_tables();
57	20					358	%next::METHOD_CACHE = ();
58	20					237	$_initialized = 1;
59							}
60
61							sub uninitialize {
62							# why bother if we don't have anything ...
63	6	50		6	1	91	return unless keys %MRO;
64	6					115	_remove_method_dispatch_tables();
65	6					66	%next::METHOD_CACHE = ();
66	6					65	$_initialized = 0;
67							}
68
69	1			1	1	13	sub reinitialize { goto &initialize }
70
71							## functions for applying C3 to classes
72
73							sub _calculate_method_dispatch_tables {
74	20			20		182	my %merge_cache;
75	20					265	foreach my $class (keys %MRO) {
76	91					992	_calculate_method_dispatch_table($class, \%merge_cache);
77							}
78							}
79
80							sub _calculate_method_dispatch_table {
81	91			91		897	my ($class, $merge_cache) = @_;
82	19			19		376	no strict 'refs';
	19					198
	19					341
83	91					1199	my @MRO = calculateMRO($class, $merge_cache);
84	91					19920	$MRO{$class} = { MRO => \@MRO };
85	91					936	my $has_overload_fallback = 0;
86	91					1006	my %methods;
87							# NOTE:
88							# we do @MRO[1 .. $#MRO] here because it
89							# makes no sense to interogate the class
90							# which you are calculating for.
91	91					1491	foreach my $local (@MRO[1 .. $#MRO]) {
92							# if overload has tagged this module to
93							# have use "fallback", then we want to
94							# grab that value
95	1					12	$has_overload_fallback = ${"${local}::()"}
	120					1515
96	120	100				1526	if defined ${"${local}::()"};
97	120					1230	foreach my $method (grep { defined &{"${local}::$_"} } keys %{"${local}::"}) {
	697					6129
	697					12025
	120					2200
98							# skip if already overriden in local class
99	109	100				928	next unless !defined *{"${class}::$method"}{CODE};
	109					1962
100	59					1090	$methods{$method} = {
101							orig => "${local}::$method",
102	77	100				1657	code => \&{"${local}::$method"}
103							} unless exists $methods{$method};
104							}
105							}
106							# now stash them in our %MRO table
107	91					4678	$MRO{$class}->{methods} = \%methods;
108	91					3049	$MRO{$class}->{has_overload_fallback} = $has_overload_fallback;
109							}
110
111							sub _apply_method_dispatch_tables {
112	20			20		240	foreach my $class (keys %MRO) {
113	91					908	_apply_method_dispatch_table($class);
114							}
115							}
116
117							sub _apply_method_dispatch_table {
118	91			91		1038	my $class = shift;
119	19			19		342	no strict 'refs';
	19					211
	19					293
120	91	100				1066	${"${class}::()"} = $MRO{$class}->{has_overload_fallback}
	1					13
121							if $MRO{$class}->{has_overload_fallback};
122	91					793	foreach my $method (keys %{$MRO{$class}->{methods}}) {
	91					1663
123	59					598	*{"${class}::$method"} = $MRO{$class}->{methods}->{$method}->{code};
	59					972
124							}
125							}
126
127							sub _remove_method_dispatch_tables {
128	6			6		89	foreach my $class (keys %MRO) {
129	31					285	_remove_method_dispatch_table($class);
130							}
131							}
132
133							sub _remove_method_dispatch_table {
134	31			31		266	my $class = shift;
135	19			19		518	no strict 'refs';
	19					225
	19					279
136	31	50				393	delete ${"${class}::"}{"()"} if $MRO{$class}->{has_overload_fallback};
	0					0
137	31					335	foreach my $method (keys %{$MRO{$class}->{methods}}) {
	31					378
138	23					325	delete ${"${class}::"}{$method}
	24					407
139	24					543	if defined *{"${class}::${method}"}{CODE} &&
140	24	100	66			1204	(*{"${class}::${method}"}{CODE} eq $MRO{$class}->{methods}->{$method}->{code});
141							}
142							}
143
144							## functions for calculating C3 MRO
145
146							sub calculateMRO {
147	129			129	1	1364	my ($class, $merge_cache) = @_;
148							return Algorithm::C3::merge($class, sub {
149	19			19		475	no strict 'refs';
	19					269
	19					1075
150	304			304		26497	@{$_[0] . '::ISA'};
	304					5713
151	129					2271	}, $merge_cache);
152							}
153
154							package # hide me from PAUSE
155							next;
156
157	19			19		675	use strict;
	19					175
	19					253
158	19			19		950	use warnings;
	19					193
	19					308
159
160	19			19		381	use Scalar::Util 'blessed';
	19					171
	19					520
161
162							our $VERSION = '0.05';
163
164							our %METHOD_CACHE;
165
166							sub method {
167	25			25		832	my $indirect = caller() =~ /^(?:next\|maybe::next)$/;
168	25	100				331	my $level = $indirect ? 2 : 1;
169
170	25					259	my ($method_caller, $label, @label);
171	25					598	while ($method_caller = (caller($level++))[3]) {
172	26					424	@label = (split '::', $method_caller);
173	26					294	$label = pop @label;
174							last unless
175	26	100	66			534	$label eq '(eval)' \|\|
176							$label eq '__ANON__';
177							}
178	25					315	my $caller = join '::' => @label;
179	25					223	my $self = $_[0];
180	25		66			799	my $class = blessed($self) \|\| $self;
181
182	25		100			496	my $method = $METHOD_CACHE{"$class\|$caller\|$label"} \|\|= do {
183
184	22					252	my @MRO = Class::C3::calculateMRO($class);
185
186	22					11708	my $current;
187	22					267	while ($current = shift @MRO) {
188	36	100				415	last if $caller eq $current;
189							}
190
191	19			19		347	no strict 'refs';
	19					212
	19					751
192	22					182	my $found;
193	22					223	foreach my $class (@MRO) {
194	36	100	100			839	next if (defined $Class::C3::MRO{$class} &&
195							defined $Class::C3::MRO{$class}{methods}{$label});
196	32	100				272	last if (defined ($found = *{$class . '::' . $label}{CODE}));
	32					496
197							}
198
199	22					292	$found;
200							};
201
202	25	100				330	return $method if $indirect;
203
204	21	100				329	die "No next::method '$label' found for $self" if !$method;
205
206	18					149	goto &{$method};
	18					308
207							}
208
209	2			2		82	sub can { method($_[0]) }
210
211							package # hide me from PAUSE
212							maybe::next;
213
214	19			19		407	use strict;
	19					205
	19					248
215	19			19		310	use warnings;
	19					176
	19					285
216
217							our $VERSION = '0.01';
218
219	2		66	2		23	sub method { (next::method($_[0]) \|\| return)->(@_) }
220
221							1;
222
223							__END__
224
225							=pod
226
227							=head1 NAME
228
229							Class::C3 - A pragma to use the C3 method resolution order algortihm
230
231							=head1 SYNOPSIS
232
233							package A;
234							use Class::C3;
235							sub hello { 'A::hello' }
236
237							package B;
238							use base 'A';
239							use Class::C3;
240
241							package C;
242							use base 'A';
243							use Class::C3;
244
245							sub hello { 'C::hello' }
246
247							package D;
248							use base ('B', 'C');
249							use Class::C3;
250
251							# Classic Diamond MI pattern
252							# <A>
253							# / \
254							# <B> <C>
255							# \ /
256							# <D>
257
258							package main;
259
260							# initializez the C3 module
261							# (formerly called in INIT)
262							Class::C3::initialize();
263
264							print join ', ' => Class::C3::calculateMRO('Diamond_D') # prints D, B, C, A
265
266							print D->hello() # prints 'C::hello' instead of the standard p5 'A::hello'
267
268							D->can('hello')->(); # can() also works correctly
269							UNIVERSAL::can('D', 'hello'); # as does UNIVERSAL::can()
270
271							=head1 DESCRIPTION
272
273							This is pragma to change Perl 5's standard method resolution order from depth-first left-to-right
274							(a.k.a - pre-order) to the more sophisticated C3 method resolution order.
275
276							=head2 What is C3?
277
278							C3 is the name of an algorithm which aims to provide a sane method resolution order under multiple
279							inheritence. It was first introduced in the langauge Dylan (see links in the L<SEE ALSO> section),
280							and then later adopted as the prefered MRO (Method Resolution Order) for the new-style classes in
281							Python 2.3. Most recently it has been adopted as the 'canonical' MRO for Perl 6 classes, and the
282							default MRO for Parrot objects as well.
283
284							=head2 How does C3 work.
285
286							C3 works by always preserving local precendence ordering. This essentially means that no class will
287							appear before any of it's subclasses. Take the classic diamond inheritence pattern for instance:
288
289							<A>
290							/ \
291							<B> <C>
292							\ /
293							<D>
294
295							The standard Perl 5 MRO would be (D, B, A, C). The result being that B<A> appears before B<C>, even
296							though B<C> is the subclass of B<A>. The C3 MRO algorithm however, produces the following MRO
297							(D, B, C, A), which does not have this same issue.
298
299							This example is fairly trival, for more complex examples and a deeper explaination, see the links in
300							the L<SEE ALSO> section.
301
302							=head2 How does this module work?
303
304							This module uses a technique similar to Perl 5's method caching. When C<Class::C3::initialize> is
305							called, this module calculates the MRO of all the classes which called C<use Class::C3>. It then
306							gathers information from the symbol tables of each of those classes, and builds a set of method
307							aliases for the correct dispatch ordering. Once all these C3-based method tables are created, it
308							then adds the method aliases into the local classes symbol table.
309
310							The end result is actually classes with pre-cached method dispatch. However, this caching does not
311							do well if you start changing your C<@ISA> or messing with class symbol tables, so you should consider
312							your classes to be effectively closed. See the L<CAVEATS> section for more details.
313
314							=head1 OPTIONAL LOWERCASE PRAGMA
315
316							This release also includes an optional module B<c3> in the F<opt/> folder. I did not include this in
317							the regular install since lowercase module names are considered I<"bad"> by some people. However I
318							think that code looks much nicer like this:
319
320							package MyClass;
321							use c3;
322
323							The the more clunky:
324
325							package MyClass;
326							use Class::C3;
327
328							But hey, it's your choice, thats why it is optional.
329
330							=head1 FUNCTIONS
331
332							=over 4
333
334							=item B<calculateMRO ($class)>
335
336							Given a C<$class> this will return an array of class names in the proper C3 method resolution order.
337
338							=item B<initialize>
339
340							This B<must be called> to initalize the C3 method dispatch tables, this module B<will not work> if
341							you do not do this. It is advised to do this as soon as possible B<after> loading any classes which
342							use C3. Here is a quick code example:
343
344							package Foo;
345							use Class::C3;
346							# ... Foo methods here
347
348							package Bar;
349							use Class::C3;
350							use base 'Foo';
351							# ... Bar methods here
352
353							package main;
354
355							Class::C3::initialize(); # now it is safe to use Foo and Bar
356
357							This function used to be called automatically for you in the INIT phase of the perl compiler, but
358							that lead to warnings if this module was required at runtime. After discussion with my user base
359							(the L<DBIx::Class> folks), we decided that calling this in INIT was more of an annoyance than a
360							convience. I apologize to anyone this causes problems for (although i would very suprised if I had
361							any other users other than the L<DBIx::Class> folks). The simplest solution of course is to define
362							your own INIT method which calls this function.
363
364							NOTE:
365
366							If C<initialize> detects that C<initialize> has already been executed, it will L</uninitialize> and
367							clear the MRO cache first.
368
369							=item B<uninitialize>
370
371							Calling this function results in the removal of all cached methods, and the restoration of the old Perl 5
372							style dispatch order (depth-first, left-to-right).
373
374							=item B<reinitialize>
375
376							This is an alias for L</initialize> above.
377
378							=back
379
380							=head1 METHOD REDISPATCHING
381
382							It is always useful to be able to re-dispatch your method call to the "next most applicable method". This
383							module provides a pseudo package along the lines of C<SUPER::> or C<NEXT::> which will re-dispatch the
384							method along the C3 linearization. This is best show with an examples.
385
386							# a classic diamond MI pattern ...
387							<A>
388							/ \
389							<B> <C>
390							\ /
391							<D>
392
393							package A;
394							use c3;
395							sub foo { 'A::foo' }
396
397							package B;
398							use base 'A';
399							use c3;
400							sub foo { 'B::foo => ' . (shift)->next::method() }
401
402							package B;
403							use base 'A';
404							use c3;
405							sub foo { 'C::foo => ' . (shift)->next::method() }
406
407							package D;
408							use base ('B', 'C');
409							use c3;
410							sub foo { 'D::foo => ' . (shift)->next::method() }
411
412							print D->foo; # prints out "D::foo => B::foo => C::foo => A::foo"
413
414							A few things to note. First, we do not require you to add on the method name to the C<next::method>
415							call (this is unlike C<NEXT::> and C<SUPER::> which do require that). This helps to enforce the rule
416							that you cannot dispatch to a method of a different name (this is how C<NEXT::> behaves as well).
417
418							The next thing to keep in mind is that you will need to pass all arguments to C<next::method> it can
419							not automatically use the current C<@_>.
420
421							If C<next::method> cannot find a next method to re-dispatch the call to, it will throw an exception.
422							You can use C<next::can> to see if C<next::method> will succeed before you call it like so:
423
424							$self->next::method(@_) if $self->next::can;
425
426							Additionally, you can use C<maybe::next::method> as a shortcut to only call the next method if it exists.
427							The previous example could be simply written as:
428
429							$self->maybe::next::method(@_);
430
431							There are some caveats about using C<next::method>, see below for those.
432
433							=head1 CAVEATS
434
435							This module used to be labeled as I<experimental>, however it has now been pretty heavily tested by
436							the good folks over at L<DBIx::Class> and I am confident this module is perfectly usable for
437							whatever your needs might be.
438
439							But there are still caveats, so here goes ...
440
441							=over 4
442
443							=item Use of C<SUPER::>.
444
445							The idea of C<SUPER::> under multiple inheritence is ambigious, and generally not recomended anyway.
446							However, it's use in conjuntion with this module is very much not recommended, and in fact very
447							discouraged. The recommended approach is to instead use the supplied C<next::method> feature, see
448							more details on it's usage above.
449
450							=item Changing C<@ISA>.
451
452							It is the author's opinion that changing C<@ISA> at runtime is pure insanity anyway. However, people
453							do it, so I must caveat. Any changes to the C<@ISA> will not be reflected in the MRO calculated by this
454							module, and therefor probably won't even show up. If you do this, you will need to call C<reinitialize>
455							in order to recalulate B<all> method dispatch tables. See the C<reinitialize> documentation and an example
456							in F<t/20_reinitialize.t> for more information.
457
458							=item Adding/deleting methods from class symbol tables.
459
460							This module calculates the MRO for each requested class by interogatting the symbol tables of said classes.
461							So any symbol table manipulation which takes place after our INIT phase is run will not be reflected in
462							the calculated MRO. Just as with changing the C<@ISA>, you will need to call C<reinitialize> for any
463							changes you make to take effect.
464
465							=item Calling C<next::method> from methods defined outside the class
466
467							There is an edge case when using C<next::method> from within a subroutine which was created in a different
468							module than the one it is called from. It sounds complicated, but it really isn't. Here is an example which
469							will not work correctly:
470
471							*Foo::foo = sub { (shift)->next::method(@_) };
472
473							The problem exists because the anonymous subroutine being assigned to the glob C<*Foo::foo> will show up
474							in the call stack as being called C<__ANON__> and not C<foo> as you might expect. Since C<next::method>
475							uses C<caller> to find the name of the method it was called in, it will fail in this case.
476
477							But fear not, there is a simple solution. The module C<Sub::Name> will reach into the perl internals and
478							assign a name to an anonymous subroutine for you. Simply do this:
479
480							use Sub::Name 'subname';
481							*Foo::foo = subname 'Foo::foo' => sub { (shift)->next::method(@_) };
482
483							and things will Just Work. Of course this is not always possible to do, but to be honest, I just can't
484							manage to find a workaround for it, so until someone gives me a working patch this will be a known
485							limitation of this module.
486
487							=back
488
489							=head1 CODE COVERAGE
490
491							I use B<Devel::Cover> to test the code coverage of my tests, below is the B<Devel::Cover> report on this
492							module's test suite.
493
494							---------------------------- ------ ------ ------ ------ ------ ------ ------
495							File stmt bran cond sub pod time total
496							---------------------------- ------ ------ ------ ------ ------ ------ ------
497							Class/C3.pm 98.3 84.4 80.0 96.2 100.0 98.4 94.4
498							---------------------------- ------ ------ ------ ------ ------ ------ ------
499							Total 98.3 84.4 80.0 96.2 100.0 98.4 94.4
500							---------------------------- ------ ------ ------ ------ ------ ------ ------
501
502							=head1 SEE ALSO
503
504							=head2 The original Dylan paper
505
506							=over 4
507
508							=item L<http://www.webcom.com/haahr/dylan/linearization-oopsla96.html>
509
510							=back
511
512							=head2 The prototype Perl 6 Object Model uses C3
513
514							=over 4
515
516							=item L<http://svn.openfoundry.org/pugs/perl5/Perl6-MetaModel/>
517
518							=back
519
520							=head2 Parrot now uses C3
521
522							=over 4
523
524							=item L<http://aspn.activestate.com/ASPN/Mail/Message/perl6-internals/2746631>
525
526							=item L<http://use.perl.org/~autrijus/journal/25768>
527
528							=back
529
530							=head2 Python 2.3 MRO related links
531
532							=over 4
533
534							=item L<http://www.python.org/2.3/mro.html>
535
536							=item L<http://www.python.org/2.2.2/descrintro.html#mro>
537
538							=back
539
540							=head2 C3 for TinyCLOS
541
542							=over 4
543
544							=item L<http://www.call-with-current-continuation.org/eggs/c3.html>
545
546							=back
547
548							=head1 ACKNOWLEGEMENTS
549
550							=over 4
551
552							=item Thanks to Matt S. Trout for using this module in his module L<DBIx::Class>
553							and finding many bugs and providing fixes.
554
555							=item Thanks to Justin Guenther for making C<next::method> more robust by handling
556							calls inside C<eval> and anon-subs.
557
558							=item Thanks to Robert Norris for adding support for C<next::can> and
559							C<maybe::next::method>.
560
561							=back
562
563							=head1 AUTHOR
564
565							Stevan Little, E<lt>stevan@iinteractive.comE<gt>
566
567							Brandon L. Black, E<lt>blblack@gmail.comE<gt>
568
569							=head1 COPYRIGHT AND LICENSE
570
571							Copyright 2005, 2006 by Infinity Interactive, Inc.
572
573							L<http://www.iinteractive.com>
574
575							This library is free software; you can redistribute it and/or modify
576							it under the same terms as Perl itself.
577
578							=cut
579