1 package DBM::Deep::Engine;
6 use warnings FATAL => 'all';
8 use DBM::Deep::Iterator ();
11 # * Every method in here assumes that the storage has been appropriately
12 # safeguarded. This can be anything from flock() to some sort of manual
13 # mutex. But, it's the caller's responsability to make sure that this has
16 sub SIG_HASH () { 'H' }
17 sub SIG_ARRAY () { 'A' }
25 This is an internal-use-only object for L<DBM::Deep/>. It mediates the low-level
26 mapping between the L<DBM::Deep/> objects and the storage medium.
28 The purpose of this documentation is to provide low-level documentation for
29 developers. It is B<not> intended to be used by the general public. This
30 documentation and what it documents can and will change without notice.
34 The engine exposes an API to the DBM::Deep objects (DBM::Deep, DBM::Deep::Array,
35 and DBM::Deep::Hash) for their use to access the actual stored values. This API
46 =item * make_reference
64 =item * lock_exclusive
72 They are explained in their own sections below. These methods, in turn, may
73 provide some bounds-checking, but primarily act to instantiate objects in the
74 Engine::Sector::* hierarchy and dispatch to them.
78 Transactions in DBM::Deep are implemented using a variant of MVCC. This attempts
79 to keep the amount of actual work done against the file low while stil providing
80 Atomicity, Consistency, and Isolation. Durability, unfortunately, cannot be done
85 If another process uses a transaction slot and writes stuff to it, then
86 terminates, the data that process wrote it still within the file. In order to
87 address this, there is also a transaction staleness counter associated within
88 every write. Each time a transaction is started, that process increments that
89 transaction's staleness counter. If, when it reads a value, the staleness
90 counters aren't identical, DBM::Deep will consider the value on disk to be stale
95 The fourth leg of ACID is Durability, the guarantee that when a commit returns,
96 the data will be there the next time you read from it. This should be regardless
97 of any crashes or powerdowns in between the commit and subsequent read.
98 DBM::Deep does provide that guarantee; once the commit returns, all of the data
99 has been transferred from the transaction shadow to the HEAD. The issue arises
100 with partial commits - a commit that is interrupted in some fashion. In keeping
101 with DBM::Deep's "tradition" of very light error-checking and non-existent
102 error-handling, there is no way to recover from a partial commit. (This is
103 probably a failure in Consistency as well as Durability.)
105 Other DBMSes use transaction logs (a separate file, generally) to achieve
106 Durability. As DBM::Deep is a single-file, we would have to do something
107 similar to what SQLite and BDB do in terms of committing using synchonized
108 writes. To do this, we would have to use a much higher RAM footprint and some
109 serious programming that make my head hurts just to think about it.
113 =head2 read_value( $obj, $key )
115 This takes an object that provides _base_offset() and a string. It returns the
116 value stored in the corresponding Sector::Value's data section.
120 sub read_value { die "read_value must be implemented in a child class" }
122 =head2 get_classname( $obj )
124 This takes an object that provides _base_offset() and returns the classname (if
125 any) associated with it.
127 It delegates to Sector::Reference::get_classname() for the heavy lifting.
129 It performs a staleness check.
133 sub get_classname { die "get_classname must be implemented in a child class" }
135 =head2 make_reference( $obj, $old_key, $new_key )
137 This takes an object that provides _base_offset() and two strings. The
138 strings correspond to the old key and new key, respectively. This operation
139 is equivalent to (given C<< $db->{foo} = []; >>) C<< $db->{bar} = $db->{foo} >>.
141 This returns nothing.
145 sub make_reference { die "make_reference must be implemented in a child class" }
147 =head2 key_exists( $obj, $key )
149 This takes an object that provides _base_offset() and a string for
150 the key to be checked. This returns 1 for true and "" for false.
154 sub key_exists { die "key_exists must be implemented in a child class" }
156 =head2 delete_key( $obj, $key )
158 This takes an object that provides _base_offset() and a string for
159 the key to be deleted. This returns the result of the Sector::Reference
164 sub delete_key { die "delete_key must be implemented in a child class" }
166 =head2 write_value( $obj, $key, $value )
168 This takes an object that provides _base_offset(), a string for the
169 key, and a value. This value can be anything storable within L<DBM::Deep/>.
171 This returns 1 upon success.
175 sub write_value { die "write_value must be implemented in a child class" }
179 This takes an object that provides _base_offset(). It will do everything needed
180 in order to properly initialize all values for necessary functioning. If this is
181 called upon an already initialized object, this will also reset the inode.
187 sub setup { die "setup must be implemented in a child class" }
189 =head2 begin_work( $obj )
191 This takes an object that provides _base_offset(). It will set up all necessary
192 bookkeeping in order to run all work within a transaction.
194 If $obj is already within a transaction, an error wiill be thrown. If there are
195 no more available transactions, an error will be thrown.
201 sub begin_work { die "begin_work must be implemented in a child class" }
203 =head2 rollback( $obj )
205 This takes an object that provides _base_offset(). It will revert all
206 actions taken within the running transaction.
208 If $obj is not within a transaction, an error will be thrown.
214 sub rollback { die "rollback must be implemented in a child class" }
216 =head2 commit( $obj )
218 This takes an object that provides _base_offset(). It will apply all
219 actions taken within the transaction to the HEAD.
221 If $obj is not within a transaction, an error will be thrown.
227 sub commit { die "commit must be implemented in a child class" }
229 =head2 get_next_key( $obj, $prev_key )
231 This takes an object that provides _base_offset() and an optional string
232 representing the prior key returned via a prior invocation of this method.
234 This method delegates to C<< DBM::Deep::Iterator->get_next_key() >>.
238 # XXX Add staleness here
241 my ($obj, $prev_key) = @_;
243 # XXX Need to add logic about resetting the iterator if any key in the
244 # reference has changed
245 unless ( $prev_key ) {
246 $obj->{iterator} = $self->iterator_class->new({
247 base_offset => $obj->_base_offset,
252 return $obj->{iterator}->get_next_key( $obj );
255 =head2 lock_exclusive()
257 This takes an object that provides _base_offset(). It will guarantee that
258 the storage has taken precautions to be safe for a write.
260 This returns nothing.
267 return $self->storage->lock_exclusive( $obj );
272 This takes an object that provides _base_offset(). It will guarantee that
273 the storage has taken precautions to be safe for a read.
275 This returns nothing.
282 return $self->storage->lock_shared( $obj );
287 This takes an object that provides _base_offset(). It will guarantee that
288 the storage has released the most recently-taken lock.
290 This returns nothing.
298 my $rv = $self->storage->unlock( $obj );
305 =head1 INTERNAL METHODS
307 The following methods are internal-use-only to DBM::Deep::Engine and its
314 This takes no arguments. It will do everything necessary to flush all things to
315 disk. This is usually called during unlock() and setup().
317 This returns nothing.
324 # Why do we need to have the storage flush? Shouldn't autoflush take care of
325 # things? -RobK, 2008-06-26
326 $self->storage->flush;
331 =head2 load_sector( $loc )
333 This takes an id/location/offset and loads the sector based on the engine's
338 sub load_sector { $_[0]->sector_type->load( @_ ) }
340 =head2 cache / clear_cache
342 This is the cache of loaded Reference sectors.
346 sub cache { $_[0]{cache} ||= {} }
347 sub clear_cache { %{$_[0]->cache} = () }
351 The following are readonly attributes.
363 sub storage { $_[0]{storage} }
365 sub sector_type { die "sector_type must be implemented in a child class" }