1DBM::Deep::Engine(3) User Contributed Perl Documentation DBM::Deep::Engine(3)
2
3
4
6 DBM::Deep::Engine - mediate mapping between DBM::Deep objects and
7 storage medium
8
10 This is an internal-use-only object for DBM::Deep. It mediates the low-
11 level mapping between the DBM::Deep objects and the storage medium.
12
13 The purpose of this documentation is to provide low-level documentation
14 for developers. It is not intended to be used by the general public.
15 This documentation and what it documents can and will change without
16 notice.
17
19 The engine exposes an API to the DBM::Deep objects (DBM::Deep,
20 DBM::Deep::Array, and DBM::Deep::Hash) for their use to access the
21 actual stored values. This API is the following:
22
23 • new
24
25 • read_value
26
27 • get_classname
28
29 • make_reference
30
31 • key_exists
32
33 • delete_key
34
35 • write_value
36
37 • get_next_key
38
39 • setup
40
41 • clear
42
43 • begin_work
44
45 • commit
46
47 • rollback
48
49 • lock_exclusive
50
51 • lock_shared
52
53 • unlock
54
55 They are explained in their own sections below. These methods, in turn,
56 may provide some bounds-checking, but primarily act to instantiate
57 objects in the Engine::Sector::* hierarchy and dispatch to them.
58
60 Transactions in DBM::Deep are implemented using a variant of MVCC. This
61 attempts to keep the amount of actual work done against the file low
62 while still providing Atomicity, Consistency, and Isolation.
63 Durability, unfortunately, cannot be done with only one file.
64
65 STALENESS
66 If another process uses a transaction slot and writes stuff to it, then
67 terminates, the data that process wrote is still within the file. In
68 order to address this, there is also a transaction staleness counter
69 associated within every write. Each time a transaction is started,
70 that process increments that transaction's staleness counter. If, when
71 it reads a value, the staleness counters aren't identical, DBM::Deep
72 will consider the value on disk to be stale and discard it.
73
74 DURABILITY
75 The fourth leg of ACID is Durability, the guarantee that when a commit
76 returns, the data will be there the next time you read from it. This
77 should be regardless of any crashes or powerdowns in between the commit
78 and subsequent read. DBM::Deep does provide that guarantee; once the
79 commit returns, all of the data has been transferred from the
80 transaction shadow to the HEAD. The issue arises with partial commits -
81 a commit that is interrupted in some fashion. In keeping with
82 DBM::Deep's "tradition" of very light error-checking and non-existent
83 error-handling, there is no way to recover from a partial commit. (This
84 is probably a failure in Consistency as well as Durability.)
85
86 Other DBMSes use transaction logs (a separate file, generally) to
87 achieve Durability. As DBM::Deep is a single-file, we would have to do
88 something similar to what SQLite and BDB do in terms of committing
89 using synchronized writes. To do this, we would have to use a much
90 higher RAM footprint and some serious programming that makes my head
91 hurt just to think about it.
92
94 read_value( $obj, $key )
95 This takes an object that provides _base_offset() and a string. It
96 returns the value stored in the corresponding Sector::Value's data
97 section.
98
99 get_classname( $obj )
100 This takes an object that provides _base_offset() and returns the
101 classname (if any) associated with it.
102
103 It delegates to Sector::Reference::get_classname() for the heavy
104 lifting.
105
106 It performs a staleness check.
107
108 make_reference( $obj, $old_key, $new_key )
109 This takes an object that provides _base_offset() and two strings. The
110 strings correspond to the old key and new key, respectively. This
111 operation is equivalent to (given "$db->{foo} = [];") "$db->{bar} =
112 $db->{foo}".
113
114 This returns nothing.
115
116 key_exists( $obj, $key )
117 This takes an object that provides _base_offset() and a string for the
118 key to be checked. This returns 1 for true and "" for false.
119
120 delete_key( $obj, $key )
121 This takes an object that provides _base_offset() and a string for the
122 key to be deleted. This returns the result of the Sector::Reference
123 delete_key() method.
124
125 write_value( $obj, $key, $value )
126 This takes an object that provides _base_offset(), a string for the
127 key, and a value. This value can be anything storable within DBM::Deep.
128
129 This returns 1 upon success.
130
131 setup( $obj )
132 This takes an object that provides _base_offset(). It will do
133 everything needed in order to properly initialize all values for
134 necessary functioning. If this is called upon an already initialized
135 object, this will also reset the inode.
136
137 This returns 1.
138
139 begin_work( $obj )
140 This takes an object that provides _base_offset(). It will set up all
141 necessary bookkeeping in order to run all work within a transaction.
142
143 If $obj is already within a transaction, an error will be thrown. If
144 there are no more available transactions, an error will be thrown.
145
146 This returns undef.
147
148 rollback( $obj )
149 This takes an object that provides _base_offset(). It will revert all
150 actions taken within the running transaction.
151
152 If $obj is not within a transaction, an error will be thrown.
153
154 This returns 1.
155
156 commit( $obj )
157 This takes an object that provides _base_offset(). It will apply all
158 actions taken within the transaction to the HEAD.
159
160 If $obj is not within a transaction, an error will be thrown.
161
162 This returns 1.
163
164 get_next_key( $obj, $prev_key )
165 This takes an object that provides _base_offset() and an optional
166 string representing the prior key returned via a prior invocation of
167 this method.
168
169 This method delegates to "DBM::Deep::Iterator->get_next_key()".
170
171 lock_exclusive()
172 This takes an object that provides _base_offset(). It will guarantee
173 that the storage has taken precautions to be safe for a write.
174
175 This returns nothing.
176
177 lock_shared()
178 This takes an object that provides _base_offset(). It will guarantee
179 that the storage has taken precautions to be safe for a read.
180
181 This returns nothing.
182
183 unlock()
184 This takes an object that provides _base_offset(). It will guarantee
185 that the storage has released the most recently-taken lock.
186
187 This returns nothing.
188
190 The following methods are internal-use-only to DBM::Deep::Engine and
191 its child classes.
192
193 flush()
194 This takes no arguments. It will do everything necessary to flush all
195 things to disk. This is usually called during unlock() and setup().
196
197 This returns nothing.
198
199 load_sector( $loc )
200 This takes an id/location/offset and loads the sector based on the
201 engine's defined sector type.
202
203 clear( $obj )
204 This takes an object that provides _base_offset() and deletes all its
205 elements, returning nothing.
206
207 cache / clear_cache
208 This is the cache of loaded Reference sectors.
209
210 supports( $option )
211 This returns a boolean depending on if this instance of DBM::Dep
212 supports that feature. $option can be one of:
213
214 • transactions
215
216 • singletons
217
218 Any other value will return false.
219
221 The following are readonly attributes.
222
223 • storage
224
225 • sector_type
226
227 • iterator_class
228
229
230
231perl v5.32.1 2021-01-27 DBM::Deep::Engine(3)