This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Hash-Util-FieldHash version bump.
[perl5.git] / ext / Hash-Util-FieldHash / lib / Hash / Util / FieldHash.pm
CommitLineData
1e73acc8
AS
1package Hash::Util::FieldHash;
2
3use 5.009004;
4use strict;
5use warnings;
1e73acc8
AS
6use Scalar::Util qw( reftype);
7
15262a79 8our $VERSION = '1.19';
3ae03d21 9
1e73acc8
AS
10require Exporter;
11our @ISA = qw(Exporter);
12our %EXPORT_TAGS = (
13 'all' => [ qw(
14 fieldhash
15 fieldhashes
d74d639b
AS
16 idhash
17 idhashes
18 id
19 id_2obj
20 register
1e73acc8
AS
21 )],
22);
23our @EXPORT_OK = ( @{ $EXPORT_TAGS{'all'} } );
1e73acc8 24
1e73acc8
AS
25{
26 require XSLoader;
6ff38c27
AS
27 my %ob_reg; # private object registry
28 sub _ob_reg { \ %ob_reg }
da4061d3 29 XSLoader::load();
1e73acc8
AS
30}
31
32sub fieldhash (\%) {
33 for ( shift ) {
34 return unless ref() && reftype( $_) eq 'HASH';
35 return $_ if Hash::Util::FieldHash::_fieldhash( $_, 0);
d74d639b
AS
36 return $_ if Hash::Util::FieldHash::_fieldhash( $_, 2) == 2;
37 return;
38 }
39}
40
41sub idhash (\%) {
42 for ( shift ) {
43 return unless ref() && reftype( $_) eq 'HASH';
44 return $_ if Hash::Util::FieldHash::_fieldhash( $_, 0);
45 return $_ if Hash::Util::FieldHash::_fieldhash( $_, 1) == 1;
1e73acc8
AS
46 return;
47 }
48}
49
50sub fieldhashes { map &fieldhash( $_), @_ }
d74d639b 51sub idhashes { map &idhash( $_), @_ }
1e73acc8
AS
52
531;
54__END__
55
56=head1 NAME
57
d74d639b 58Hash::Util::FieldHash - Support for Inside-Out Classes
1e73acc8 59
b299c47c 60=head1 SYNOPSIS
1e73acc8 61
d74d639b 62 ### Create fieldhashes
1e73acc8 63 use Hash::Util qw(fieldhash fieldhashes);
6ff38c27 64
1e73acc8
AS
65 # Create a single field hash
66 fieldhash my %foo;
6ff38c27 67
1e73acc8
AS
68 # Create three at once...
69 fieldhashes \ my(%foo, %bar, %baz);
70 # ...or any number
71 fieldhashes @hashrefs;
72
d74d639b 73 ### Create an idhash and register it for garbage collection
ada3dff8 74 use Hash::Util::FieldHash qw(idhash register);
d74d639b 75 idhash my %name;
ada3dff8 76 my $object = \ do { my $o };
d74d639b 77 # register the idhash for garbage collection with $object
ada3dff8
RGS
78 register($object, \ %name);
79 # the following entry will be deleted when $object goes out of scope
80 $name{$object} = 'John Doe';
b299c47c 81
d74d639b 82 ### Register an ordinary hash for garbage collection
ada3dff8 83 use Hash::Util::FieldHash qw(id register);
d74d639b 84 my %name;
ada3dff8
RGS
85 my $object = \ do { my $o };
86 # register the hash %name for garbage collection of $object's id
87 register $object, \ %name;
88 # the following entry will be deleted when $object goes out of scope
89 $name{id $object} = 'John Doe';
d74d639b 90
b299c47c 91=head1 FUNCTIONS
1e73acc8 92
d74d639b 93C<Hash::Util::FieldHash> offers a number of functions in support of
b299c47c 94L<The Inside-out Technique> of class construction.
1e73acc8
AS
95
96=over
97
d74d639b 98=item id
1e73acc8 99
d74d639b 100 id($obj)
1e73acc8 101
d74d639b
AS
102Returns the reference address of a reference $obj. If $obj is
103not a reference, returns $obj.
1e73acc8 104
d74d639b 105This function is a stand-in replacement for
712d8896 106L<Scalar::Util::refaddr|Scalar::Util/refaddr>,
b79fa73c 107that is, it returns
d74d639b
AS
108the reference address of its argument as a numeric value. The only
109difference is that C<refaddr()> returns C<undef> when given a
110non-reference while C<id()> returns its argument unchanged.
1e73acc8 111
d74d639b
AS
112C<id()> also uses a caching technique that makes it faster when
113the id of an object is requested often, but slower if it is needed
114only once or twice.
1e73acc8 115
d74d639b 116=item id_2obj
1e73acc8 117
d74d639b 118 $obj = id_2obj($id)
1e73acc8 119
d74d639b
AS
120If C<$id> is the id of a registered object (see L</register>), returns
121the object, otherwise an undefined value. For registered objects this
122is the inverse function of C<id()>.
1e73acc8 123
d74d639b 124=item register
1e73acc8 125
d74d639b
AS
126 register($obj)
127 register($obj, @hashrefs)
1e73acc8 128
d74d639b
AS
129In the first form, registers an object to work with for the function
130C<id_2obj()>. In the second form, it additionally marks the given
131hashrefs down for garbage collection. This means that when the object
132goes out of scope, any entries in the given hashes under the key of
133C<id($obj)> will be deleted from the hashes.
1e73acc8 134
d74d639b
AS
135It is a fatal error to register a non-reference $obj. Any non-hashrefs
136among the following arguments are silently ignored.
1e73acc8 137
d74d639b
AS
138It is I<not> an error to register the same object multiple times with
139varying sets of hashrefs. Any hashrefs that are not registered yet
140will be added, others ignored.
1e73acc8 141
d74d639b
AS
142Registry also implies thread support. When a new thread is created,
143all references are replaced with new ones, including all objects.
144If a hash uses the reference address of an object as a key, that
145connection would be broken. With a registered object, its id will
146be updated in all hashes registered with it.
1e73acc8 147
d74d639b 148=item idhash
1e73acc8 149
d74d639b 150 idhash my %hash
1e73acc8 151
d74d639b 152Makes an idhash from the argument, which must be a hash.
1e73acc8 153
d74d639b
AS
154An I<idhash> works like a normal hash, except that it stringifies a
155I<reference used as a key> differently. A reference is stringified
156as if the C<id()> function had been invoked on it, that is, its
157reference address in decimal is used as the key.
1e73acc8 158
d74d639b 159=item idhashes
1e73acc8 160
ada3dff8 161 idhashes \ my(%hash, %gnash, %trash)
d74d639b 162 idhashes \ @hashrefs
1e73acc8 163
d74d639b
AS
164Creates many idhashes from its hashref arguments. Returns those
165arguments that could be converted or their number in scalar context.
1e73acc8 166
d74d639b 167=item fieldhash
1e73acc8 168
d74d639b 169 fieldhash %hash;
1e73acc8 170
d74d639b
AS
171Creates a single fieldhash. The argument must be a hash. Returns
172a reference to the given hash if successful, otherwise nothing.
1e73acc8 173
d74d639b
AS
174A I<fieldhash> is, in short, an idhash with auto-registry. When an
175object (or, indeed, any reference) is used as a fieldhash key, the
176fieldhash is automatically registered for garbage collection with
177the object, as if C<register $obj, \ %fieldhash> had been called.
1e73acc8 178
d74d639b 179=item fieldhashes
1e73acc8 180
d74d639b 181 fieldhashes @hashrefs;
1e73acc8 182
d74d639b
AS
183Creates any number of field hashes. Arguments must be hash references.
184Returns the converted hashrefs in list context, their number in scalar
185context.
1e73acc8
AS
186
187=back
188
b299c47c 189=head1 DESCRIPTION
1e73acc8 190
d74d639b
AS
191A word on terminology: I shall use the term I<field> for a scalar
192piece of data that a class associates with an object. Other terms that
193have been used for this concept are "object variable", "(object) property",
ada3dff8 194"(object) attribute" and more. Especially "attribute" has some currency
d74d639b
AS
195among Perl programmer, but that clashes with the C<attributes> pragma. The
196term "field" also has some currency in this sense and doesn't seem
197to conflict with other Perl terminology.
198
199In Perl, an object is a blessed reference. The standard way of associating
b7ffd429 200data with an object is to store the data inside the object's body, that is,
d74d639b
AS
201the piece of data pointed to by the reference.
202
203In consequence, if two or more classes want to access an object they
ada3dff8 204I<must> agree on the type of reference and also on the organization of
d74d639b
AS
205data within the object body. Failure to agree on the type results in
206immediate death when the wrong method tries to access an object. Failure
207to agree on data organization may lead to one class trampling over the
208data of another.
209
210This object model leads to a tight coupling between subclasses.
211If one class wants to inherit from another (and both classes access
212object data), the classes must agree about implementation details.
213Inheritance can only be used among classes that are maintained together,
214in a single source or not.
215
216In particular, it is not possible to write general-purpose classes
217in this technique, classes that can advertise themselves as "Put me
218on your @ISA list and use my methods". If the other class has different
219ideas about how the object body is used, there is trouble.
220
69895b01 221For reference C<Name_hash> in L</Example 1> shows the standard implementation of
d74d639b
AS
222a simple class C<Name> in the well-known hash based way. It also demonstrates
223the predictable failure to construct a common subclass C<NamedFile>
224of C<Name> and the class C<IO::File> (whose objects I<must> be globrefs).
225
226Thus, techniques are of interest that store object data I<not> in
227the object body but some other place.
228
229=head2 The Inside-out Technique
230
231With I<inside-out> classes, each class declares a (typically lexical)
232hash for each field it wants to use. The reference address of an
233object is used as the hash key. By definition, the reference address
234is unique to each object so this guarantees a place for each field that
69895b01
KW
235is private to the class and unique to each object. See C<Name_id>
236in L</Example 1> for a simple example.
d74d639b
AS
237
238In comparison to the standard implementation where the object is a
239hash and the fields correspond to hash keys, here the fields correspond
240to hashes, and the object determines the hash key. Thus the hashes
241appear to be turned I<inside out>.
242
243The body of an object is never examined by an inside-out class, only
244its reference address is used. This allows for the body of an actual
245object to be I<anything at all> while the object methods of the class
246still work as designed. This is a key feature of inside-out classes.
247
248=head2 Problems of Inside-out
249
250Inside-out classes give us freedom of inheritance, but as usual there
251is a price.
252
253Most obviously, there is the necessity of retrieving the reference
254address of an object for each data access. It's a minor inconvenience,
255but it does clutter the code.
256
257More important (and less obvious) is the necessity of garbage
258collection. When a normal object dies, anything stored in the
259object body is garbage-collected by perl. With inside-out objects,
260Perl knows nothing about the data stored in field hashes by a class,
261but these must be deleted when the object goes out of scope. Thus
262the class must provide a C<DESTROY> method to take care of that.
263
264In the presence of multiple classes it can be non-trivial
265to make sure that every relevant destructor is called for
266every object. Perl calls the first one it finds on the
267inheritance tree (if any) and that's it.
268
269A related issue is thread-safety. When a new thread is created,
270the Perl interpreter is cloned, which implies that all reference
271addresses in use will be replaced with new ones. Thus, if a class
272tries to access a field of a cloned object its (cloned) data will
273still be stored under the now invalid reference address of the
274original in the parent thread. A general C<CLONE> method must
275be provided to re-establish the association.
276
277=head2 Solutions
278
279C<Hash::Util::FieldHash> addresses these issues on several
280levels.
281
282The C<id()> function is provided in addition to the
283existing C<Scalar::Util::refaddr()>. Besides its short name
284it can be a little faster under some circumstances (and a
285bit slower under others). Benchmark if it matters. The
286working of C<id()> also allows the use of the class name
b299c47c 287as a I<generic object> as described L<further down|/"The Generic Object">.
d74d639b
AS
288
289The C<id()> function is incorporated in I<id hashes> in the sense
290that it is called automatically on every key that is used with
291the hash. No explicit call is necessary.
292
293The problems of garbage collection and thread safety are both
294addressed by the function C<register()>. It registers an object
295together with any number of hashes. Registry means that when the
296object dies, an entry in any of the hashes under the reference
297address of this object will be deleted. This guarantees garbage
298collection in these hashes. It also means that on thread
299cloning the object's entries in registered hashes will be
300replaced with updated entries whose key is the cloned object's
301reference address. Thus the object-data association becomes
302thread-safe.
303
304Object registry is best done when the object is initialized
305for use with a class. That way, garbage collection and thread
306safety are established for every object and every field that is
307initialized.
308
309Finally, I<field hashes> incorporate all these functions in one
310package. Besides automatically calling the C<id()> function
311on every object used as a key, the object is registered with
312the field hash on first use. Classes based on field hashes
313are fully garbage-collected and thread safe without further
314measures.
315
316=head2 More Problems
317
318Another problem that occurs with inside-out classes is serialization.
319Since the object data is not in its usual place, standard routines
320like C<Storable::freeze()>, C<Storable::thaw()> and
321C<Data::Dumper::Dumper()> can't deal with it on their own. Both
322C<Data::Dumper> and C<Storable> provide the necessary hooks to
323make things work, but the functions or methods used by the hooks
324must be provided by each inside-out class.
325
326A general solution to the serialization problem would require another
92a34c45 327level of registry, one that associates I<classes> and fields.
b7ffd429 328So far, the functions of C<Hash::Util::FieldHash> are unaware of
d74d639b
AS
329any classes, which I consider a feature. Therefore C<Hash::Util::FieldHash>
330doesn't address the serialization problems.
331
332=head2 The Generic Object
333
334Classes based on the C<id()> function (and hence classes based on
335C<idhash()> and C<fieldhash()>) show a peculiar behavior in that
336the class name can be used like an object. Specifically, methods
337that set or read data associated with an object continue to work as
338class methods, just as if the class name were an object, distinct from
339all other objects, with its own data. This object may be called
340the I<generic object> of the class.
341
ada3dff8 342This works because field hashes respond to keys that are not references
d74d639b
AS
343like a normal hash would and use the string offered as the hash key.
344Thus, if a method is called as a class method, the field hash is presented
345with the class name instead of an object and blithely uses it as a key.
346Since the keys of real objects are decimal numbers, there is no
347conflict and the slot in the field hash can be used like any other.
348The C<id()> function behaves correspondingly with respect to non-reference
349arguments.
350
351Two possible uses (besides ignoring the property) come to mind.
352A singleton class could be implemented this using the generic object.
353If necessary, an C<init()> method could die or ignore calls with
354actual objects (references), so only the generic object will ever exist.
355
356Another use of the generic object would be as a template. It is
357a convenient place to store class-specific defaults for various
358fields to be used in actual object initialization.
359
360Usually, the feature can be entirely ignored. Calling I<object
361methods> as I<class methods> normally leads to an error and isn't used
362routinely anywhere. It may be a problem that this error isn't
363indicated by a class with a generic object.
364
365=head2 How to use Field Hashes
1e73acc8
AS
366
367Traditionally, the definition of an inside-out class contains a bare
368block inside which a number of lexical hashes are declared and the
369basic accessor methods defined, usually through C<Scalar::Util::refaddr>.
370Further methods may be defined outside this block. There has to be
371a DESTROY method and, for thread support, a CLONE method.
372
ada3dff8 373When field hashes are used, the basic structure remains the same.
1e73acc8
AS
374Each lexical hash will be made a field hash. The call to C<refaddr>
375can be omitted from the accessor methods. DESTROY and CLONE methods
376are not necessary.
377
378If you have an existing inside-out class, simply making all hashes
379field hashes with no other change should make no difference. Through
380the calls to C<refaddr> or equivalent, the field hashes never get to
381see a reference and work like normal hashes. Your DESTROY (and
382CLONE) methods are still needed.
383
384To make the field hashes kick in, it is easiest to redefine C<refaddr>
385as
386
387 sub refaddr { shift }
388
389instead of importing it from C<Scalar::Util>. It should now be possible
390to disable DESTROY and CLONE. Note that while it isn't disabled,
391DESTROY will be called before the garbage collection of field hashes,
6ff38c27
AS
392so it will be invoked with a functional object and will continue to
393function.
394
395It is not desirable to import the functions C<fieldhash> and/or
396C<fieldhashes> into every class that is going to use them. They
397are only used once to set up the class. When the class is up and running,
398these functions serve no more purpose.
1e73acc8 399
1e73acc8
AS
400If there are only a few field hashes to declare, it is simplest to
401
402 use Hash::Util::FieldHash;
403
404early and call the functions qualified:
405
406 Hash::Util::FieldHash::fieldhash my %foo;
407
408Otherwise, import the functions into a convenient package like
d74d639b 409C<HUF> or, more general, C<Aux>
1e73acc8
AS
410
411 {
412 package Aux;
413 use Hash::Util::FieldHash ':all';
414 }
415
416and call
417
418 Aux::fieldhash my %foo;
419
420as needed.
421
d74d639b
AS
422=head2 Garbage-Collected Hashes
423
424Garbage collection in a field hash means that entries will "spontaneously"
425disappear when the object that created them disappears. That must be
426borne in mind, especially when looping over a field hash. If anything
427you do inside the loop could cause an object to go out of scope, a
428random key may be deleted from the hash you are looping over. That
429can throw the loop iterator, so it's best to cache a consistent snapshot
430of the keys and/or values and loop over that. You will still have to
431check that a cached entry still exists when you get to it.
432
433Garbage collection can be confusing when keys are created in a field hash
434from normal scalars as well as references. Once a reference is I<used> with
435a field hash, the entry will be collected, even if it was later overwritten
436with a plain scalar key (every positive integer is a candidate). This
437is true even if the original entry was deleted in the meantime. In fact,
438deletion from a field hash, and also a test for existence constitute
439I<use> in this sense and create a liability to delete the entry when
440the reference goes out of scope. If you happen to create an entry
441with an identical key from a string or integer, that will be collected
442instead. Thus, mixed use of references and plain scalars as field hash
443keys is not entirely supported.
444
b299c47c 445=head1 EXAMPLES
1e73acc8 446
d74d639b
AS
447The examples show a very simple class that implements a I<name>, consisting
448of a first and last name (no middle initial). The name class has four
449methods:
1e73acc8 450
d74d639b 451=over
1e73acc8 452
d74d639b 453=item * C<init()>
1e73acc8 454
d74d639b
AS
455An object method that initializes the first and last name to its
456two arguments. If called as a class method, C<init()> creates an
457object in the given class and initializes that.
1e73acc8 458
d74d639b 459=item * C<first()>
1e73acc8 460
d74d639b 461Retrieve the first name
1e73acc8 462
d74d639b 463=item * C<last()>
1e73acc8 464
d74d639b 465Retrieve the last name
1e73acc8 466
d74d639b 467=item * C<name()>
1e73acc8 468
d74d639b 469Retrieve the full name, the first and last name joined by a blank.
1e73acc8 470
d74d639b 471=back
1e73acc8 472
d74d639b
AS
473The examples show this class implemented with different levels of
474support by C<Hash::Util::FieldHash>. All supported combinations
475are shown. The difference between implementations is often quite
476small. The implementations are:
1e73acc8 477
d74d639b 478=over
1e73acc8 479
d74d639b 480=item * C<Name_hash>
1e73acc8 481
d74d639b
AS
482A conventional (not inside-out) implementation where an object is
483a hash that stores the field values, without support by
484C<Hash::Util::FieldHash>. This implementation doesn't allow
485arbitrary inheritance.
486
487=item * C<Name_id>
488
489Inside-out implementation based on the C<id()> function. It needs
490a C<DESTROY> method. For thread support a C<CLONE> method (not shown)
491would also be needed. Instead of C<Hash::Util::FieldHash::id()> the
492function C<Scalar::Util::refaddr> could be used with very little
493functional difference. This is the basic pattern of an inside-out
494class.
495
496=item * C<Name_idhash>
497
69895b01 498Idhash-based inside-out implementation. Like C<Name_id> it needs
d74d639b
AS
499a C<DESTROY> method and would need C<CLONE> for thread support.
500
501=item * C<Name_id_reg>
502
503Inside-out implementation based on the C<id()> function with explicit
504object registry. No destructor is needed and objects are thread safe.
505
506=item * C<Name_idhash_reg>
507
508Idhash-based inside-out implementation with explicit object registry.
509No destructor is needed and objects are thread safe.
510
511=item * C<Name_fieldhash>
512
b7ffd429 513FieldHash-based inside-out implementation. Object registry happens
d74d639b
AS
514automatically. No destructor is needed and objects are thread safe.
515
516=back
517
518These examples are realized in the code below, which could be copied
519to a file F<Example.pm>.
520
521=head2 Example 1
522
523 use strict; use warnings;
524
525 {
555bd962
BG
526 package Name_hash; # standard implementation: the
527 # object is a hash
d74d639b
AS
528 sub init {
529 my $obj = shift;
ada3dff8 530 my ($first, $last) = @_;
d74d639b
AS
531 # create an object if called as class method
532 $obj = bless {}, $obj unless ref $obj;
533 $obj->{ first} = $first;
534 $obj->{ last} = $last;
535 $obj;
536 }
537
538 sub first { shift()->{ first} }
539 sub last { shift()->{ last} }
540
541 sub name {
542 my $n = shift;
543 join ' ' => $n->first, $n->last;
544 }
6ff38c27 545
1e73acc8
AS
546 }
547
d74d639b
AS
548 {
549 package Name_id;
ada3dff8 550 use Hash::Util::FieldHash qw(id);
d74d639b 551
ada3dff8 552 my (%first, %last);
d74d639b
AS
553
554 sub init {
555 my $obj = shift;
ada3dff8 556 my ($first, $last) = @_;
d74d639b
AS
557 # create an object if called as class method
558 $obj = bless \ my $o, $obj unless ref $obj;
559 $first{ id $obj} = $first;
560 $last{ id $obj} = $last;
561 $obj;
562 }
563
564 sub first { $first{ id shift()} }
565 sub last { $last{ id shift()} }
566
567 sub name {
568 my $n = shift;
569 join ' ' => $n->first, $n->last;
570 }
571
572 sub DESTROY {
573 my $id = id shift;
574 delete $first{ $id};
575 delete $last{ $id};
576 }
1e73acc8 577
1e73acc8
AS
578 }
579
d74d639b
AS
580 {
581 package Name_idhash;
582 use Hash::Util::FieldHash;
583
ada3dff8 584 Hash::Util::FieldHash::idhashes( \ my (%first, %last) );
d74d639b
AS
585
586 sub init {
587 my $obj = shift;
ada3dff8 588 my ($first, $last) = @_;
d74d639b
AS
589 # create an object if called as class method
590 $obj = bless \ my $o, $obj unless ref $obj;
591 $first{ $obj} = $first;
592 $last{ $obj} = $last;
593 $obj;
594 }
595
596 sub first { $first{ shift()} }
597 sub last { $last{ shift()} }
598
599 sub name {
600 my $n = shift;
601 join ' ' => $n->first, $n->last;
602 }
603
604 sub DESTROY {
605 my $n = shift;
606 delete $first{ $n};
607 delete $last{ $n};
608 }
609
1e73acc8
AS
610 }
611
d74d639b
AS
612 {
613 package Name_id_reg;
ada3dff8 614 use Hash::Util::FieldHash qw(id register);
d74d639b 615
ada3dff8 616 my (%first, %last);
d74d639b
AS
617
618 sub init {
619 my $obj = shift;
ada3dff8 620 my ($first, $last) = @_;
d74d639b
AS
621 # create an object if called as class method
622 $obj = bless \ my $o, $obj unless ref $obj;
ada3dff8 623 register( $obj, \ (%first, %last) );
d74d639b
AS
624 $first{ id $obj} = $first;
625 $last{ id $obj} = $last;
626 $obj;
627 }
628
629 sub first { $first{ id shift()} }
630 sub last { $last{ id shift()} }
631
632 sub name {
633 my $n = shift;
634 join ' ' => $n->first, $n->last;
635 }
636 }
637
638 {
639 package Name_idhash_reg;
ada3dff8 640 use Hash::Util::FieldHash qw(register);
d74d639b 641
ada3dff8 642 Hash::Util::FieldHash::idhashes \ my (%first, %last);
d74d639b
AS
643
644 sub init {
645 my $obj = shift;
ada3dff8 646 my ($first, $last) = @_;
d74d639b
AS
647 # create an object if called as class method
648 $obj = bless \ my $o, $obj unless ref $obj;
ada3dff8 649 register( $obj, \ (%first, %last) );
d74d639b
AS
650 $first{ $obj} = $first;
651 $last{ $obj} = $last;
652 $obj;
653 }
654
655 sub first { $first{ shift()} }
656 sub last { $last{ shift()} }
657
658 sub name {
659 my $n = shift;
660 join ' ' => $n->first, $n->last;
661 }
662 }
663
664 {
665 package Name_fieldhash;
666 use Hash::Util::FieldHash;
667
ada3dff8 668 Hash::Util::FieldHash::fieldhashes \ my (%first, %last);
d74d639b
AS
669
670 sub init {
671 my $obj = shift;
ada3dff8 672 my ($first, $last) = @_;
d74d639b
AS
673 # create an object if called as class method
674 $obj = bless \ my $o, $obj unless ref $obj;
675 $first{ $obj} = $first;
676 $last{ $obj} = $last;
677 $obj;
678 }
679
680 sub first { $first{ shift()} }
681 sub last { $last{ shift()} }
682
683 sub name {
684 my $n = shift;
685 join ' ' => $n->first, $n->last;
686 }
687 }
688
689 1;
690
b299c47c 691To exercise the various implementations the script L<below|/"Example 2"> can
d74d639b
AS
692be used.
693
694It sets up a class C<Name> that is a mirror of one of the implementation
695classes C<Name_hash>, C<Name_id>, ..., C<Name_fieldhash>. That determines
696which implementation is run.
697
698The script first verifies the function of the C<Name> class.
699
b7ffd429 700In the second step, the free inheritability of the implementation
d74d639b
AS
701(or lack thereof) is demonstrated. For this purpose it constructs
702a class called C<NamedFile> which is a common subclass of C<Name> and
703the standard class C<IO::File>. This puts inheritability to the test
704because objects of C<IO::File> I<must> be globrefs. Objects of C<NamedFile>
705should behave like a file opened for reading and also support the C<name()>
706method. This class juncture works with exception of the C<Name_hash>
707implementation, where object initialization fails because of the
708incompatibility of object bodies.
709
710=head2 Example 2
711
712 use strict; use warnings; $| = 1;
713
714 use Example;
715
716 {
717 package Name;
9f21ab77 718 use parent 'Name_id'; # define here which implementation to run
d74d639b
AS
719 }
720
721
722 # Verify that the base package works
ada3dff8 723 my $n = Name->init(qw(Albert Einstein));
d74d639b
AS
724 print $n->name, "\n";
725 print "\n";
726
727 # Create a named file handle (See definition below)
ada3dff8 728 my $nf = NamedFile->init(qw(/tmp/x Filomena File));
d74d639b
AS
729 # use as a file handle...
730 for ( 1 .. 3 ) {
731 my $l = <$nf>;
732 print "line $_: $l";
733 }
734 # ...and as a Name object
735 print "...brought to you by ", $nf->name, "\n";
736 exit;
737
738
739 # Definition of NamedFile
740 package NamedFile;
9f21ab77
RS
741 use parent 'Name';
742 use parent 'IO::File';
d74d639b
AS
743
744 sub init {
745 my $obj = shift;
ada3dff8 746 my ($file, $first, $last) = @_;
d74d639b 747 $obj = $obj->IO::File::new() unless ref $obj;
ada3dff8
RGS
748 $obj->open($file) or die "Can't read '$file': $!";
749 $obj->Name::init($first, $last);
d74d639b
AS
750 }
751 __END__
1e73acc8 752
1e73acc8 753
b299c47c 754=head1 GUTS
1e73acc8
AS
755
756To make C<Hash::Util::FieldHash> work, there were two changes to
b7b1e41b 757F<perl> itself. C<PERL_MAGIC_uvar> was made available for hashes,
1e73acc8
AS
758and weak references now call uvar C<get> magic after a weakref has been
759cleared. The first feature is used to make field hashes intercept
760their keys upon access. The second one triggers garbage collection.
761
762=head2 The C<PERL_MAGIC_uvar> interface for hashes
763
764C<PERL_MAGIC_uvar> I<get> magic is called from C<hv_fetch_common> and
765C<hv_delete_common> through the function C<hv_magic_uvar_xkey>, which
766defines the interface. The call happens for hashes with "uvar" magic
767if the C<ufuncs> structure has equal values in the C<uf_val> and C<uf_set>
768fields. Hashes are unaffected if (and as long as) these fields
769hold different values.
770
771Upon the call, the C<mg_obj> field will hold the hash key to be accessed.
772Upon return, the C<SV*> value in C<mg_obj> will be used in place of the
773original key in the hash access. The integer index value in the first
774parameter will be the C<action> value from C<hv_fetch_common>, or -1
775if the call is from C<hv_delete_common>.
776
777This is a template for a function suitable for the C<uf_val> field in
778a C<ufuncs> structure for this call. The C<uf_set> and C<uf_index>
779fields are irrelevant.
780
781 IV watch_key(pTHX_ IV action, SV* field) {
782 MAGIC* mg = mg_find(field, PERL_MAGIC_uvar);
783 SV* keysv = mg->mg_obj;
784 /* Do whatever you need to. If you decide to
785 supply a different key newkey, return it like this
786 */
787 sv_2mortal(newkey);
788 mg->mg_obj = newkey;
789 return 0;
790 }
791
792=head2 Weakrefs call uvar magic
793
794When a weak reference is stored in an C<SV> that has "uvar" magic, C<set>
795magic is called after the reference has gone stale. This hook can be
796used to trigger further garbage-collection activities associated with
797the referenced object.
798
799=head2 How field hashes work
800
801The three features of key hashes, I<key replacement>, I<thread support>,
802and I<garbage collection> are supported by a data structure called
6ff38c27
AS
803the I<object registry>. This is a private hash where every object
804is stored. An "object" in this sense is any reference (blessed or
805unblessed) that has been used as a field hash key.
1e73acc8
AS
806
807The object registry keeps track of references that have been used as
808field hash keys. The keys are generated from the reference address
809like in a field hash (though the registry isn't a field hash). Each
810value is a weak copy of the original reference, stored in an C<SV> that
811is itself magical (C<PERL_MAGIC_uvar> again). The magical structure
812holds a list (another hash, really) of field hashes that the reference
813has been used with. When the weakref becomes stale, the magic is
814activated and uses the list to delete the reference from all field
815hashes it has been used with. After that, the entry is removed from
816the object registry itself. Implicitly, that frees the magic structure
817and the storage it has been using.
818
819Whenever a reference is used as a field hash key, the object registry
820is checked and a new entry is made if necessary. The field hash is
821then added to the list of fields this reference has used.
822
823The object registry is also used to repair a field hash after thread
824cloning. Here, the entire object registry is processed. For every
825reference found there, the field hashes it has used are visited and
826the entry is updated.
827
828=head2 Internal function Hash::Util::FieldHash::_fieldhash
829
830 # test if %hash is a field hash
831 my $result = _fieldhash \ %hash, 0;
832
833 # make %hash a field hash
834 my $result = _fieldhash \ %hash, 1;
835
836C<_fieldhash> is the internal function used to create field hashes.
837It takes two arguments, a hashref and a mode. If the mode is boolean
838false, the hash is not changed but tested if it is a field hash. If
839the hash isn't a field hash the return value is boolean false. If it
840is, the return value indicates the mode of field hash. When called with
841a boolean true mode, it turns the given hash into a field hash of this
842mode, returning the mode of the created field hash. C<_fieldhash>
843does not erase the given hash.
844
845Currently there is only one type of field hash, and only the boolean
846value of the mode makes a difference, but that may change.
847
848=head1 AUTHOR
849
3ae03d21
AS
850Anno Siegel (ANNO) wrote the xs code and the changes in perl proper
851Jerry Hedden (JDHEDDEN) made it faster
1e73acc8
AS
852
853=head1 COPYRIGHT AND LICENSE
854
b299c47c 855Copyright (C) 2006-2007 by (Anno Siegel)
1e73acc8
AS
856
857This library is free software; you can redistribute it and/or modify
858it under the same terms as Perl itself, either Perl version 5.8.7 or,
859at your option, any later version of Perl 5 you may have available.
860
861=cut