| 1 |
.\" -*- mode: troff; coding: utf-8 -*- |
| 2 |
.\" Automatically generated by Pod::Man 5.0102 (Pod::Simple 3.45) |
| 3 |
.\" |
| 4 |
.\" Standard preamble: |
| 5 |
.\" ======================================================================== |
| 6 |
.de Sp \" Vertical space (when we can't use .PP) |
| 7 |
.if t .sp .5v |
| 8 |
.if n .sp |
| 9 |
.. |
| 10 |
.de Vb \" Begin verbatim text |
| 11 |
.ft CW |
| 12 |
.nf |
| 13 |
.ne \\$1 |
| 14 |
.. |
| 15 |
.de Ve \" End verbatim text |
| 16 |
.ft R |
| 17 |
.fi |
| 18 |
.. |
| 19 |
.\" \*(C` and \*(C' are quotes in nroff, nothing in troff, for use with C<>. |
| 20 |
.ie n \{\ |
| 21 |
. ds C` "" |
| 22 |
. ds C' "" |
| 23 |
'br\} |
| 24 |
.el\{\ |
| 25 |
. ds C` |
| 26 |
. ds C' |
| 27 |
'br\} |
| 28 |
.\" |
| 29 |
.\" Escape single quotes in literal strings from groff's Unicode transform. |
| 30 |
.ie \n(.g .ds Aq \(aq |
| 31 |
.el .ds Aq ' |
| 32 |
.\" |
| 33 |
.\" If the F register is >0, we'll generate index entries on stderr for |
| 34 |
.\" titles (.TH), headers (.SH), subsections (.SS), items (.Ip), and index |
| 35 |
.\" entries marked with X<> in POD. Of course, you'll have to process the |
| 36 |
.\" output yourself in some meaningful fashion. |
| 37 |
.\" |
| 38 |
.\" Avoid warning from groff about undefined register 'F'. |
| 39 |
.de IX |
| 40 |
.. |
| 41 |
.nr rF 0 |
| 42 |
.if \n(.g .if rF .nr rF 1 |
| 43 |
.if (\n(rF:(\n(.g==0)) \{\ |
| 44 |
. if \nF \{\ |
| 45 |
. de IX |
| 46 |
. tm Index:\\$1\t\\n%\t"\\$2" |
| 47 |
.. |
| 48 |
. if !\nF==2 \{\ |
| 49 |
. nr % 0 |
| 50 |
. nr F 2 |
| 51 |
. \} |
| 52 |
. \} |
| 53 |
.\} |
| 54 |
.rr rF |
| 55 |
.\" ======================================================================== |
| 56 |
.\" |
| 57 |
.IX Title "Storable 3perl" |
| 58 |
.TH Storable 3perl 2025-11-16 "perl v5.40.1" "Perl Programmers Reference Guide" |
| 59 |
.\" For nroff, turn off justification. Always turn off hyphenation; it makes |
| 60 |
.\" way too many mistakes in technical documents. |
| 61 |
.if n .ad l |
| 62 |
.nh |
| 63 |
.SH NAME |
| 64 |
Storable \- persistence for Perl data structures |
| 65 |
.SH SYNOPSIS |
| 66 |
.IX Header "SYNOPSIS" |
| 67 |
.Vb 3 |
| 68 |
\& use Storable; |
| 69 |
\& store \e%table, \*(Aqfile\*(Aq; |
| 70 |
\& $hashref = retrieve(\*(Aqfile\*(Aq); |
| 71 |
\& |
| 72 |
\& use Storable qw(nstore store_fd nstore_fd freeze thaw dclone); |
| 73 |
\& |
| 74 |
\& # Network order |
| 75 |
\& nstore \e%table, \*(Aqfile\*(Aq; |
| 76 |
\& $hashref = retrieve(\*(Aqfile\*(Aq); # There is NO nretrieve() |
| 77 |
\& |
| 78 |
\& # Storing to and retrieving from an already opened file |
| 79 |
\& store_fd \e@array, \e*STDOUT; |
| 80 |
\& nstore_fd \e%table, \e*STDOUT; |
| 81 |
\& $aryref = fd_retrieve(\e*SOCKET); |
| 82 |
\& $hashref = fd_retrieve(\e*SOCKET); |
| 83 |
\& |
| 84 |
\& # Serializing to memory |
| 85 |
\& $serialized = freeze \e%table; |
| 86 |
\& %table_clone = %{ thaw($serialized) }; |
| 87 |
\& |
| 88 |
\& # Deep (recursive) cloning |
| 89 |
\& $cloneref = dclone($ref); |
| 90 |
\& |
| 91 |
\& # Advisory locking |
| 92 |
\& use Storable qw(lock_store lock_nstore lock_retrieve) |
| 93 |
\& lock_store \e%table, \*(Aqfile\*(Aq; |
| 94 |
\& lock_nstore \e%table, \*(Aqfile\*(Aq; |
| 95 |
\& $hashref = lock_retrieve(\*(Aqfile\*(Aq); |
| 96 |
.Ve |
| 97 |
.SH DESCRIPTION |
| 98 |
.IX Header "DESCRIPTION" |
| 99 |
The Storable package brings persistence to your Perl data structures |
| 100 |
containing SCALAR, ARRAY, HASH or REF objects, i.e. anything that can be |
| 101 |
conveniently stored to disk and retrieved at a later time. |
| 102 |
.PP |
| 103 |
It can be used in the regular procedural way by calling \f(CW\*(C`store\*(C'\fR with |
| 104 |
a reference to the object to be stored, along with the file name where |
| 105 |
the image should be written. |
| 106 |
.PP |
| 107 |
The routine returns \f(CW\*(C`undef\*(C'\fR for I/O problems or other internal error, |
| 108 |
a true value otherwise. Serious errors are propagated as a \f(CW\*(C`die\*(C'\fR exception. |
| 109 |
.PP |
| 110 |
To retrieve data stored to disk, use \f(CW\*(C`retrieve\*(C'\fR with a file name. |
| 111 |
The objects stored into that file are recreated into memory for you, |
| 112 |
and a \fIreference\fR to the root object is returned. In case an I/O error |
| 113 |
occurs while reading, \f(CW\*(C`undef\*(C'\fR is returned instead. Other serious |
| 114 |
errors are propagated via \f(CW\*(C`die\*(C'\fR. |
| 115 |
.PP |
| 116 |
Since storage is performed recursively, you might want to stuff references |
| 117 |
to objects that share a lot of common data into a single array or hash |
| 118 |
table, and then store that object. That way, when you retrieve back the |
| 119 |
whole thing, the objects will continue to share what they originally shared. |
| 120 |
.PP |
| 121 |
At the cost of a slight header overhead, you may store to an already |
| 122 |
opened file descriptor using the \f(CW\*(C`store_fd\*(C'\fR routine, and retrieve |
| 123 |
from a file via \f(CW\*(C`fd_retrieve\*(C'\fR. Those names aren't imported by default, |
| 124 |
so you will have to do that explicitly if you need those routines. |
| 125 |
The file descriptor you supply must be already opened, for read |
| 126 |
if you're going to retrieve and for write if you wish to store. |
| 127 |
.PP |
| 128 |
.Vb 2 |
| 129 |
\& store_fd(\e%table, *STDOUT) || die "can\*(Aqt store to stdout\en"; |
| 130 |
\& $hashref = fd_retrieve(*STDIN); |
| 131 |
.Ve |
| 132 |
.PP |
| 133 |
You can also store data in network order to allow easy sharing across |
| 134 |
multiple platforms, or when storing on a socket known to be remotely |
| 135 |
connected. The routines to call have an initial \f(CW\*(C`n\*(C'\fR prefix for \fInetwork\fR, |
| 136 |
as in \f(CW\*(C`nstore\*(C'\fR and \f(CW\*(C`nstore_fd\*(C'\fR. At retrieval time, your data will be |
| 137 |
correctly restored so you don't have to know whether you're restoring |
| 138 |
from native or network ordered data. Double values are stored stringified |
| 139 |
to ensure portability as well, at the slight risk of loosing some precision |
| 140 |
in the last decimals. |
| 141 |
.PP |
| 142 |
When using \f(CW\*(C`fd_retrieve\*(C'\fR, objects are retrieved in sequence, one |
| 143 |
object (i.e. one recursive tree) per associated \f(CW\*(C`store_fd\*(C'\fR. |
| 144 |
.PP |
| 145 |
If you're more from the object-oriented camp, you can inherit from |
| 146 |
Storable and directly store your objects by invoking \f(CW\*(C`store\*(C'\fR as |
| 147 |
a method. The fact that the root of the to-be-stored tree is a |
| 148 |
blessed reference (i.e. an object) is special-cased so that the |
| 149 |
retrieve does not provide a reference to that object but rather the |
| 150 |
blessed object reference itself. (Otherwise, you'd get a reference |
| 151 |
to that blessed object). |
| 152 |
.SH "MEMORY STORE" |
| 153 |
.IX Header "MEMORY STORE" |
| 154 |
The Storable engine can also store data into a Perl scalar instead, to |
| 155 |
later retrieve them. This is mainly used to freeze a complex structure in |
| 156 |
some safe compact memory place (where it can possibly be sent to another |
| 157 |
process via some IPC, since freezing the structure also serializes it in |
| 158 |
effect). Later on, and maybe somewhere else, you can thaw the Perl scalar |
| 159 |
out and recreate the original complex structure in memory. |
| 160 |
.PP |
| 161 |
Surprisingly, the routines to be called are named \f(CW\*(C`freeze\*(C'\fR and \f(CW\*(C`thaw\*(C'\fR. |
| 162 |
If you wish to send out the frozen scalar to another machine, use |
| 163 |
\&\f(CW\*(C`nfreeze\*(C'\fR instead to get a portable image. |
| 164 |
.PP |
| 165 |
Note that freezing an object structure and immediately thawing it |
| 166 |
actually achieves a deep cloning of that structure: |
| 167 |
.PP |
| 168 |
.Vb 1 |
| 169 |
\& dclone(.) = thaw(freeze(.)) |
| 170 |
.Ve |
| 171 |
.PP |
| 172 |
Storable provides you with a \f(CW\*(C`dclone\*(C'\fR interface which does not create |
| 173 |
that intermediary scalar but instead freezes the structure in some |
| 174 |
internal memory space and then immediately thaws it out. |
| 175 |
.SH "ADVISORY LOCKING" |
| 176 |
.IX Header "ADVISORY LOCKING" |
| 177 |
The \f(CW\*(C`lock_store\*(C'\fR and \f(CW\*(C`lock_nstore\*(C'\fR routine are equivalent to |
| 178 |
\&\f(CW\*(C`store\*(C'\fR and \f(CW\*(C`nstore\*(C'\fR, except that they get an exclusive lock on |
| 179 |
the file before writing. Likewise, \f(CW\*(C`lock_retrieve\*(C'\fR does the same |
| 180 |
as \f(CW\*(C`retrieve\*(C'\fR, but also gets a shared lock on the file before reading. |
| 181 |
.PP |
| 182 |
As with any advisory locking scheme, the protection only works if you |
| 183 |
systematically use \f(CW\*(C`lock_store\*(C'\fR and \f(CW\*(C`lock_retrieve\*(C'\fR. If one side of |
| 184 |
your application uses \f(CW\*(C`store\*(C'\fR whilst the other uses \f(CW\*(C`lock_retrieve\*(C'\fR, |
| 185 |
you will get no protection at all. |
| 186 |
.PP |
| 187 |
The internal advisory locking is implemented using Perl's \fBflock()\fR |
| 188 |
routine. If your system does not support any form of \fBflock()\fR, or if |
| 189 |
you share your files across NFS, you might wish to use other forms |
| 190 |
of locking by using modules such as LockFile::Simple which lock a |
| 191 |
file using a filesystem entry, instead of locking the file descriptor. |
| 192 |
.SH SPEED |
| 193 |
.IX Header "SPEED" |
| 194 |
The heart of Storable is written in C for decent speed. Extra low-level |
| 195 |
optimizations have been made when manipulating perl internals, to |
| 196 |
sacrifice encapsulation for the benefit of greater speed. |
| 197 |
.SH "CANONICAL REPRESENTATION" |
| 198 |
.IX Header "CANONICAL REPRESENTATION" |
| 199 |
Normally, Storable stores elements of hashes in the order they are |
| 200 |
stored internally by Perl, i.e. pseudo-randomly. If you set |
| 201 |
\&\f(CW$Storable::canonical\fR to some \f(CW\*(C`TRUE\*(C'\fR value, Storable will store |
| 202 |
hashes with the elements sorted by their key. This allows you to |
| 203 |
compare data structures by comparing their frozen representations (or |
| 204 |
even the compressed frozen representations), which can be useful for |
| 205 |
creating lookup tables for complicated queries. |
| 206 |
.PP |
| 207 |
Canonical order does not imply network order; those are two orthogonal |
| 208 |
settings. |
| 209 |
.SH "CODE REFERENCES" |
| 210 |
.IX Header "CODE REFERENCES" |
| 211 |
Since Storable version 2.05, CODE references may be serialized with |
| 212 |
the help of B::Deparse. To enable this feature, set |
| 213 |
\&\f(CW$Storable::Deparse\fR to a true value. To enable deserialization, |
| 214 |
\&\f(CW$Storable::Eval\fR should be set to a true value. Be aware that |
| 215 |
deserialization is done through \f(CW\*(C`eval\*(C'\fR, which is dangerous if the |
| 216 |
Storable file contains malicious data. You can set \f(CW$Storable::Eval\fR |
| 217 |
to a subroutine reference which would be used instead of \f(CW\*(C`eval\*(C'\fR. See |
| 218 |
below for an example using a Safe compartment for deserialization |
| 219 |
of CODE references. |
| 220 |
.PP |
| 221 |
If \f(CW$Storable::Deparse\fR and/or \f(CW$Storable::Eval\fR are set to false |
| 222 |
values, then the value of \f(CW$Storable::forgive_me\fR (see below) is |
| 223 |
respected while serializing and deserializing. |
| 224 |
.SH "FORWARD COMPATIBILITY" |
| 225 |
.IX Header "FORWARD COMPATIBILITY" |
| 226 |
This release of Storable can be used on a newer version of Perl to |
| 227 |
serialize data which is not supported by earlier Perls. By default, |
| 228 |
Storable will attempt to do the right thing, by \f(CWcroak()\fRing if it |
| 229 |
encounters data that it cannot deserialize. However, the defaults |
| 230 |
can be changed as follows: |
| 231 |
.IP "utf8 data" 4 |
| 232 |
.IX Item "utf8 data" |
| 233 |
Perl 5.6 added support for Unicode characters with code points > 255, |
| 234 |
and Perl 5.8 has full support for Unicode characters in hash keys. |
| 235 |
Perl internally encodes strings with these characters using utf8, and |
| 236 |
Storable serializes them as utf8. By default, if an older version of |
| 237 |
Perl encounters a utf8 value it cannot represent, it will \f(CWcroak()\fR. |
| 238 |
To change this behaviour so that Storable deserializes utf8 encoded |
| 239 |
values as the string of bytes (effectively dropping the \fIis_utf8\fR flag) |
| 240 |
set \f(CW$Storable::drop_utf8\fR to some \f(CW\*(C`TRUE\*(C'\fR value. This is a form of |
| 241 |
data loss, because with \f(CW$drop_utf8\fR true, it becomes impossible to tell |
| 242 |
whether the original data was the Unicode string, or a series of bytes |
| 243 |
that happen to be valid utf8. |
| 244 |
.IP "restricted hashes" 4 |
| 245 |
.IX Item "restricted hashes" |
| 246 |
Perl 5.8 adds support for restricted hashes, which have keys |
| 247 |
restricted to a given set, and can have values locked to be read only. |
| 248 |
By default, when Storable encounters a restricted hash on a perl |
| 249 |
that doesn't support them, it will deserialize it as a normal hash, |
| 250 |
silently discarding any placeholder keys and leaving the keys and |
| 251 |
all values unlocked. To make Storable \f(CWcroak()\fR instead, set |
| 252 |
\&\f(CW$Storable::downgrade_restricted\fR to a \f(CW\*(C`FALSE\*(C'\fR value. To restore |
| 253 |
the default set it back to some \f(CW\*(C`TRUE\*(C'\fR value. |
| 254 |
.Sp |
| 255 |
The cperl PERL_PERTURB_KEYS_TOP hash strategy has a known problem with |
| 256 |
restricted hashes. |
| 257 |
.IP "huge objects" 4 |
| 258 |
.IX Item "huge objects" |
| 259 |
On 64bit systems some data structures may exceed the 2G (i.e. I32_MAX) |
| 260 |
limit. On 32bit systems also strings between I32 and U32 (2G\-4G). |
| 261 |
Since Storable 3.00 (not in perl5 core) we are able to store and |
| 262 |
retrieve these objects, even if perl5 itself is not able to handle |
| 263 |
them. These are strings longer then 4G, arrays with more then 2G |
| 264 |
elements and hashes with more then 2G elements. cperl forbids hashes |
| 265 |
with more than 2G elements, but this fail in cperl then. perl5 itself |
| 266 |
at least until 5.26 allows it, but cannot iterate over them. |
| 267 |
Note that creating those objects might cause out of memory |
| 268 |
exceptions by the operating system before perl has a chance to abort. |
| 269 |
.IP "files from future versions of Storable" 4 |
| 270 |
.IX Item "files from future versions of Storable" |
| 271 |
Earlier versions of Storable would immediately croak if they encountered |
| 272 |
a file with a higher internal version number than the reading Storable |
| 273 |
knew about. Internal version numbers are increased each time new data |
| 274 |
types (such as restricted hashes) are added to the vocabulary of the file |
| 275 |
format. This meant that a newer Storable module had no way of writing a |
| 276 |
file readable by an older Storable, even if the writer didn't store newer |
| 277 |
data types. |
| 278 |
.Sp |
| 279 |
This version of Storable will defer croaking until it encounters a data |
| 280 |
type in the file that it does not recognize. This means that it will |
| 281 |
continue to read files generated by newer Storable modules which are careful |
| 282 |
in what they write out, making it easier to upgrade Storable modules in a |
| 283 |
mixed environment. |
| 284 |
.Sp |
| 285 |
The old behaviour of immediate croaking can be re-instated by setting |
| 286 |
\&\f(CW$Storable::accept_future_minor\fR to some \f(CW\*(C`FALSE\*(C'\fR value. |
| 287 |
.PP |
| 288 |
All these variables have no effect on a newer Perl which supports the |
| 289 |
relevant feature. |
| 290 |
.SH "ERROR REPORTING" |
| 291 |
.IX Header "ERROR REPORTING" |
| 292 |
Storable uses the "exception" paradigm, in that it does not try to |
| 293 |
workaround failures: if something bad happens, an exception is |
| 294 |
generated from the caller's perspective (see Carp and \f(CWcroak()\fR). |
| 295 |
Use eval {} to trap those exceptions. |
| 296 |
.PP |
| 297 |
When Storable croaks, it tries to report the error via the \f(CWlogcroak()\fR |
| 298 |
routine from the \f(CW\*(C`Log::Agent\*(C'\fR package, if it is available. |
| 299 |
.PP |
| 300 |
Normal errors are reported by having \fBstore()\fR or \fBretrieve()\fR return \f(CW\*(C`undef\*(C'\fR. |
| 301 |
Such errors are usually I/O errors (or truncated stream errors at retrieval). |
| 302 |
.PP |
| 303 |
When Storable throws the "Max. recursion depth with nested structures |
| 304 |
exceeded" error we are already out of stack space. Unfortunately on |
| 305 |
some earlier perl versions cleaning up a recursive data structure |
| 306 |
recurses into the free calls, which will lead to stack overflows in |
| 307 |
the cleanup. This data structure is not properly cleaned up then, it |
| 308 |
will only be destroyed during global destruction. |
| 309 |
.SH "WIZARDS ONLY" |
| 310 |
.IX Header "WIZARDS ONLY" |
| 311 |
.SS Hooks |
| 312 |
.IX Subsection "Hooks" |
| 313 |
Any class may define hooks that will be called during the serialization |
| 314 |
and deserialization process on objects that are instances of that class. |
| 315 |
Those hooks can redefine the way serialization is performed (and therefore, |
| 316 |
how the symmetrical deserialization should be conducted). |
| 317 |
.PP |
| 318 |
Since we said earlier: |
| 319 |
.PP |
| 320 |
.Vb 1 |
| 321 |
\& dclone(.) = thaw(freeze(.)) |
| 322 |
.Ve |
| 323 |
.PP |
| 324 |
everything we say about hooks should also hold for deep cloning. However, |
| 325 |
hooks get to know whether the operation is a mere serialization, or a cloning. |
| 326 |
.PP |
| 327 |
Therefore, when serializing hooks are involved, |
| 328 |
.PP |
| 329 |
.Vb 1 |
| 330 |
\& dclone(.) <> thaw(freeze(.)) |
| 331 |
.Ve |
| 332 |
.PP |
| 333 |
Well, you could keep them in sync, but there's no guarantee it will always |
| 334 |
hold on classes somebody else wrote. Besides, there is little to gain in |
| 335 |
doing so: a serializing hook could keep only one attribute of an object, |
| 336 |
which is probably not what should happen during a deep cloning of that |
| 337 |
same object. |
| 338 |
.PP |
| 339 |
Here is the hooking interface: |
| 340 |
.ie n .IP """STORABLE_freeze"" \fIobj\fR, \fIcloning\fR" 4 |
| 341 |
.el .IP "\f(CWSTORABLE_freeze\fR \fIobj\fR, \fIcloning\fR" 4 |
| 342 |
.IX Item "STORABLE_freeze obj, cloning" |
| 343 |
The serializing hook, called on the object during serialization. It can be |
| 344 |
inherited, or defined in the class itself, like any other method. |
| 345 |
.Sp |
| 346 |
Arguments: \fIobj\fR is the object to serialize, \fIcloning\fR is a flag indicating |
| 347 |
whether we're in a \fBdclone()\fR or a regular serialization via \fBstore()\fR or \fBfreeze()\fR. |
| 348 |
.Sp |
| 349 |
Returned value: A LIST \f(CW\*(C`($serialized, $ref1, $ref2, ...)\*(C'\fR where \f(CW$serialized\fR |
| 350 |
is the serialized form to be used, and the optional \f(CW$ref1\fR, \f(CW$ref2\fR, etc... are |
| 351 |
extra references that you wish to let the Storable engine serialize. |
| 352 |
.Sp |
| 353 |
At deserialization time, you will be given back the same LIST, but all the |
| 354 |
extra references will be pointing into the deserialized structure. |
| 355 |
.Sp |
| 356 |
The \fBfirst time\fR the hook is hit in a serialization flow, you may have it |
| 357 |
return an empty list. That will signal the Storable engine to further |
| 358 |
discard that hook for this class and to therefore revert to the default |
| 359 |
serialization of the underlying Perl data. The hook will again be normally |
| 360 |
processed in the next serialization. |
| 361 |
.Sp |
| 362 |
Unless you know better, serializing hook should always say: |
| 363 |
.Sp |
| 364 |
.Vb 5 |
| 365 |
\& sub STORABLE_freeze { |
| 366 |
\& my ($self, $cloning) = @_; |
| 367 |
\& return if $cloning; # Regular default serialization |
| 368 |
\& .... |
| 369 |
\& } |
| 370 |
.Ve |
| 371 |
.Sp |
| 372 |
in order to keep reasonable \fBdclone()\fR semantics. |
| 373 |
.ie n .IP """STORABLE_thaw"" \fIobj\fR, \fIcloning\fR, \fIserialized\fR, ..." 4 |
| 374 |
.el .IP "\f(CWSTORABLE_thaw\fR \fIobj\fR, \fIcloning\fR, \fIserialized\fR, ..." 4 |
| 375 |
.IX Item "STORABLE_thaw obj, cloning, serialized, ..." |
| 376 |
The deserializing hook called on the object during deserialization. |
| 377 |
But wait: if we're deserializing, there's no object yet... right? |
| 378 |
.Sp |
| 379 |
Wrong: the Storable engine creates an empty one for you. If you know Eiffel, |
| 380 |
you can view \f(CW\*(C`STORABLE_thaw\*(C'\fR as an alternate creation routine. |
| 381 |
.Sp |
| 382 |
This means the hook can be inherited like any other method, and that |
| 383 |
\&\fIobj\fR is your blessed reference for this particular instance. |
| 384 |
.Sp |
| 385 |
The other arguments should look familiar if you know \f(CW\*(C`STORABLE_freeze\*(C'\fR: |
| 386 |
\&\fIcloning\fR is true when we're part of a deep clone operation, \fIserialized\fR |
| 387 |
is the serialized string you returned to the engine in \f(CW\*(C`STORABLE_freeze\*(C'\fR, |
| 388 |
and there may be an optional list of references, in the same order you gave |
| 389 |
them at serialization time, pointing to the deserialized objects (which |
| 390 |
have been processed courtesy of the Storable engine). |
| 391 |
.Sp |
| 392 |
When the Storable engine does not find any \f(CW\*(C`STORABLE_thaw\*(C'\fR hook routine, |
| 393 |
it tries to load the class by requiring the package dynamically (using |
| 394 |
the blessed package name), and then re-attempts the lookup. If at that |
| 395 |
time the hook cannot be located, the engine croaks. Note that this mechanism |
| 396 |
will fail if you define several classes in the same file, but perlmod |
| 397 |
warned you. |
| 398 |
.Sp |
| 399 |
It is up to you to use this information to populate \fIobj\fR the way you want. |
| 400 |
.Sp |
| 401 |
Returned value: none. |
| 402 |
.ie n .IP """STORABLE_attach"" \fIclass\fR, \fIcloning\fR, \fIserialized\fR" 4 |
| 403 |
.el .IP "\f(CWSTORABLE_attach\fR \fIclass\fR, \fIcloning\fR, \fIserialized\fR" 4 |
| 404 |
.IX Item "STORABLE_attach class, cloning, serialized" |
| 405 |
While \f(CW\*(C`STORABLE_freeze\*(C'\fR and \f(CW\*(C`STORABLE_thaw\*(C'\fR are useful for classes where |
| 406 |
each instance is independent, this mechanism has difficulty (or is |
| 407 |
incompatible) with objects that exist as common process-level or |
| 408 |
system-level resources, such as singleton objects, database pools, caches |
| 409 |
or memoized objects. |
| 410 |
.Sp |
| 411 |
The alternative \f(CW\*(C`STORABLE_attach\*(C'\fR method provides a solution for these |
| 412 |
shared objects. Instead of \f(CW\*(C`STORABLE_freeze\*(C'\fR \-\-> \f(CW\*(C`STORABLE_thaw\*(C'\fR, |
| 413 |
you implement \f(CW\*(C`STORABLE_freeze\*(C'\fR \-\-> \f(CW\*(C`STORABLE_attach\*(C'\fR instead. |
| 414 |
.Sp |
| 415 |
Arguments: \fIclass\fR is the class we are attaching to, \fIcloning\fR is a flag |
| 416 |
indicating whether we're in a \fBdclone()\fR or a regular de-serialization via |
| 417 |
\&\fBthaw()\fR, and \fIserialized\fR is the stored string for the resource object. |
| 418 |
.Sp |
| 419 |
Because these resource objects are considered to be owned by the entire |
| 420 |
process/system, and not the "property" of whatever is being serialized, |
| 421 |
no references underneath the object should be included in the serialized |
| 422 |
string. Thus, in any class that implements \f(CW\*(C`STORABLE_attach\*(C'\fR, the |
| 423 |
\&\f(CW\*(C`STORABLE_freeze\*(C'\fR method cannot return any references, and \f(CW\*(C`Storable\*(C'\fR |
| 424 |
will throw an error if \f(CW\*(C`STORABLE_freeze\*(C'\fR tries to return references. |
| 425 |
.Sp |
| 426 |
All information required to "attach" back to the shared resource object |
| 427 |
\&\fBmust\fR be contained \fBonly\fR in the \f(CW\*(C`STORABLE_freeze\*(C'\fR return string. |
| 428 |
Otherwise, \f(CW\*(C`STORABLE_freeze\*(C'\fR behaves as normal for \f(CW\*(C`STORABLE_attach\*(C'\fR |
| 429 |
classes. |
| 430 |
.Sp |
| 431 |
Because \f(CW\*(C`STORABLE_attach\*(C'\fR is passed the class (rather than an object), |
| 432 |
it also returns the object directly, rather than modifying the passed |
| 433 |
object. |
| 434 |
.Sp |
| 435 |
Returned value: object of type \f(CW\*(C`class\*(C'\fR |
| 436 |
.SS Predicates |
| 437 |
.IX Subsection "Predicates" |
| 438 |
Predicates are not exportable. They must be called by explicitly prefixing |
| 439 |
them with the Storable package name. |
| 440 |
.ie n .IP """Storable::last_op_in_netorder""" 4 |
| 441 |
.el .IP \f(CWStorable::last_op_in_netorder\fR 4 |
| 442 |
.IX Item "Storable::last_op_in_netorder" |
| 443 |
The \f(CWStorable::last_op_in_netorder()\fR predicate will tell you whether |
| 444 |
network order was used in the last store or retrieve operation. If you |
| 445 |
don't know how to use this, just forget about it. |
| 446 |
.ie n .IP """Storable::is_storing""" 4 |
| 447 |
.el .IP \f(CWStorable::is_storing\fR 4 |
| 448 |
.IX Item "Storable::is_storing" |
| 449 |
Returns true if within a store operation (via STORABLE_freeze hook). |
| 450 |
.ie n .IP """Storable::is_retrieving""" 4 |
| 451 |
.el .IP \f(CWStorable::is_retrieving\fR 4 |
| 452 |
.IX Item "Storable::is_retrieving" |
| 453 |
Returns true if within a retrieve operation (via STORABLE_thaw hook). |
| 454 |
.SS Recursion |
| 455 |
.IX Subsection "Recursion" |
| 456 |
With hooks comes the ability to recurse back to the Storable engine. |
| 457 |
Indeed, hooks are regular Perl code, and Storable is convenient when |
| 458 |
it comes to serializing and deserializing things, so why not use it |
| 459 |
to handle the serialization string? |
| 460 |
.PP |
| 461 |
There are a few things you need to know, however: |
| 462 |
.IP \(bu 4 |
| 463 |
From Storable 3.05 to 3.13 we probed for the stack recursion limit for references, |
| 464 |
arrays and hashes to a maximal depth of ~1200\-35000, otherwise we might |
| 465 |
fall into a stack-overflow. On JSON::XS this limit is 512 btw. With |
| 466 |
references not immediately referencing each other there's no such |
| 467 |
limit yet, so you might fall into such a stack-overflow segfault. |
| 468 |
.Sp |
| 469 |
This probing and the checks we performed have some limitations: |
| 470 |
.RS 4 |
| 471 |
.IP \(bu 4 |
| 472 |
the stack size at build time might be different at run time, eg. the |
| 473 |
stack size may have been modified with \fBulimit\fR\|(1). If it's larger at |
| 474 |
run time Storable may fail the \fBfreeze()\fR or \fBthaw()\fR unnecessarily. If |
| 475 |
it's larger at build time Storable may segmentation fault when |
| 476 |
processing a deep structure at run time. |
| 477 |
.IP \(bu 4 |
| 478 |
the stack size might be different in a thread. |
| 479 |
.IP \(bu 4 |
| 480 |
array and hash recursion limits are checked separately against the |
| 481 |
same recursion depth, a frozen structure with a large sequence of |
| 482 |
nested arrays within many nested hashes may exhaust the processor |
| 483 |
stack without triggering Storable's recursion protection. |
| 484 |
.RE |
| 485 |
.RS 4 |
| 486 |
.Sp |
| 487 |
So these now have simple defaults rather than probing at build-time. |
| 488 |
.Sp |
| 489 |
You can control the maximum array and hash recursion depths by |
| 490 |
modifying \f(CW$Storable::recursion_limit\fR and |
| 491 |
\&\f(CW$Storable::recursion_limit_hash\fR respectively. Either can be set to |
| 492 |
\&\f(CW\-1\fR to prevent any depth checks, though this isn't recommended. |
| 493 |
.Sp |
| 494 |
If you want to test what the limits are, the \fIstacksize\fR tool is |
| 495 |
included in the \f(CW\*(C`Storable\*(C'\fR distribution. |
| 496 |
.RE |
| 497 |
.IP \(bu 4 |
| 498 |
You can create endless loops if the things you serialize via \fBfreeze()\fR |
| 499 |
(for instance) point back to the object we're trying to serialize in |
| 500 |
the hook. |
| 501 |
.IP \(bu 4 |
| 502 |
Shared references among objects will not stay shared: if we're serializing |
| 503 |
the list of object [A, C] where both object A and C refer to the SAME object |
| 504 |
B, and if there is a serializing hook in A that says freeze(B), then when |
| 505 |
deserializing, we'll get [A', C'] where A' refers to B', but C' refers to D, |
| 506 |
a deep clone of B'. The topology was not preserved. |
| 507 |
.IP \(bu 4 |
| 508 |
The maximal stack recursion limit for your system is returned by |
| 509 |
\&\f(CWstack_depth()\fR and \f(CWstack_depth_hash()\fR. The hash limit is usually |
| 510 |
half the size of the array and ref limit, as the Perl hash API is not optimal. |
| 511 |
.PP |
| 512 |
That's why \f(CW\*(C`STORABLE_freeze\*(C'\fR lets you provide a list of references |
| 513 |
to serialize. The engine guarantees that those will be serialized in the |
| 514 |
same context as the other objects, and therefore that shared objects will |
| 515 |
stay shared. |
| 516 |
.PP |
| 517 |
In the above [A, C] example, the \f(CW\*(C`STORABLE_freeze\*(C'\fR hook could return: |
| 518 |
.PP |
| 519 |
.Vb 1 |
| 520 |
\& ("something", $self\->{B}) |
| 521 |
.Ve |
| 522 |
.PP |
| 523 |
and the B part would be serialized by the engine. In \f(CW\*(C`STORABLE_thaw\*(C'\fR, you |
| 524 |
would get back the reference to the B' object, deserialized for you. |
| 525 |
.PP |
| 526 |
Therefore, recursion should normally be avoided, but is nonetheless supported. |
| 527 |
.SS "Deep Cloning" |
| 528 |
.IX Subsection "Deep Cloning" |
| 529 |
There is a Clone module available on CPAN which implements deep cloning |
| 530 |
natively, i.e. without freezing to memory and thawing the result. It is |
| 531 |
aimed to replace Storable's \fBdclone()\fR some day. However, it does not currently |
| 532 |
support Storable hooks to redefine the way deep cloning is performed. |
| 533 |
.SH "Storable magic" |
| 534 |
.IX Header "Storable magic" |
| 535 |
Yes, there's a lot of that :\-) But more precisely, in UNIX systems |
| 536 |
there's a utility called \f(CW\*(C`file\*(C'\fR, which recognizes data files based on |
| 537 |
their contents (usually their first few bytes). For this to work, |
| 538 |
a certain file called \fImagic\fR needs to taught about the \fIsignature\fR |
| 539 |
of the data. Where that configuration file lives depends on the UNIX |
| 540 |
flavour; often it's something like \fI/usr/share/misc/magic\fR or |
| 541 |
\&\fI/etc/magic\fR. Your system administrator needs to do the updating of |
| 542 |
the \fImagic\fR file. The necessary signature information is output to |
| 543 |
STDOUT by invoking \fBStorable::show_file_magic()\fR. Note that the GNU |
| 544 |
implementation of the \f(CW\*(C`file\*(C'\fR utility, version 3.38 or later, |
| 545 |
is expected to contain support for recognising Storable files |
| 546 |
out-of-the-box, in addition to other kinds of Perl files. |
| 547 |
.PP |
| 548 |
You can also use the following functions to extract the file header |
| 549 |
information from Storable images: |
| 550 |
.ie n .IP "$info = Storable::file_magic( $filename )" 4 |
| 551 |
.el .IP "\f(CW$info\fR = Storable::file_magic( \f(CW$filename\fR )" 4 |
| 552 |
.IX Item "$info = Storable::file_magic( $filename )" |
| 553 |
If the given file is a Storable image return a hash describing it. If |
| 554 |
the file is readable, but not a Storable image return \f(CW\*(C`undef\*(C'\fR. If |
| 555 |
the file does not exist or is unreadable then croak. |
| 556 |
.Sp |
| 557 |
The hash returned has the following elements: |
| 558 |
.RS 4 |
| 559 |
.ie n .IP """version""" 4 |
| 560 |
.el .IP \f(CWversion\fR 4 |
| 561 |
.IX Item "version" |
| 562 |
This returns the file format version. It is a string like "2.7". |
| 563 |
.Sp |
| 564 |
Note that this version number is not the same as the version number of |
| 565 |
the Storable module itself. For instance Storable v0.7 create files |
| 566 |
in format v2.0 and Storable v2.15 create files in format v2.7. The |
| 567 |
file format version number only increment when additional features |
| 568 |
that would confuse older versions of the module are added. |
| 569 |
.Sp |
| 570 |
Files older than v2.0 will have the one of the version numbers "\-1", |
| 571 |
"0" or "1". No minor number was used at that time. |
| 572 |
.ie n .IP """version_nv""" 4 |
| 573 |
.el .IP \f(CWversion_nv\fR 4 |
| 574 |
.IX Item "version_nv" |
| 575 |
This returns the file format version as number. It is a string like |
| 576 |
"2.007". This value is suitable for numeric comparisons. |
| 577 |
.Sp |
| 578 |
The constant function \f(CW\*(C`Storable::BIN_VERSION_NV\*(C'\fR returns a comparable |
| 579 |
number that represents the highest file version number that this |
| 580 |
version of Storable fully supports (but see discussion of |
| 581 |
\&\f(CW$Storable::accept_future_minor\fR above). The constant |
| 582 |
\&\f(CW\*(C`Storable::BIN_WRITE_VERSION_NV\*(C'\fR function returns what file version |
| 583 |
is written and might be less than \f(CW\*(C`Storable::BIN_VERSION_NV\*(C'\fR in some |
| 584 |
configurations. |
| 585 |
.ie n .IP """major"", ""minor""" 4 |
| 586 |
.el .IP "\f(CWmajor\fR, \f(CWminor\fR" 4 |
| 587 |
.IX Item "major, minor" |
| 588 |
This also returns the file format version. If the version is "2.7" |
| 589 |
then major would be 2 and minor would be 7. The minor element is |
| 590 |
missing for when major is less than 2. |
| 591 |
.ie n .IP """hdrsize""" 4 |
| 592 |
.el .IP \f(CWhdrsize\fR 4 |
| 593 |
.IX Item "hdrsize" |
| 594 |
The is the number of bytes that the Storable header occupies. |
| 595 |
.ie n .IP """netorder""" 4 |
| 596 |
.el .IP \f(CWnetorder\fR 4 |
| 597 |
.IX Item "netorder" |
| 598 |
This is TRUE if the image store data in network order. This means |
| 599 |
that it was created with \fBnstore()\fR or similar. |
| 600 |
.ie n .IP """byteorder""" 4 |
| 601 |
.el .IP \f(CWbyteorder\fR 4 |
| 602 |
.IX Item "byteorder" |
| 603 |
This is only present when \f(CW\*(C`netorder\*(C'\fR is FALSE. It is the |
| 604 |
\&\f(CW$Config\fR{byteorder} string of the perl that created this image. It is |
| 605 |
a string like "1234" (32 bit little endian) or "87654321" (64 bit big |
| 606 |
endian). This must match the current perl for the image to be |
| 607 |
readable by Storable. |
| 608 |
.ie n .IP """intsize"", ""longsize"", ""ptrsize"", ""nvsize""" 4 |
| 609 |
.el .IP "\f(CWintsize\fR, \f(CWlongsize\fR, \f(CWptrsize\fR, \f(CWnvsize\fR" 4 |
| 610 |
.IX Item "intsize, longsize, ptrsize, nvsize" |
| 611 |
These are only present when \f(CW\*(C`netorder\*(C'\fR is FALSE. These are the sizes of |
| 612 |
various C datatypes of the perl that created this image. These must |
| 613 |
match the current perl for the image to be readable by Storable. |
| 614 |
.Sp |
| 615 |
The \f(CW\*(C`nvsize\*(C'\fR element is only present for file format v2.2 and |
| 616 |
higher. |
| 617 |
.ie n .IP """file""" 4 |
| 618 |
.el .IP \f(CWfile\fR 4 |
| 619 |
.IX Item "file" |
| 620 |
The name of the file. |
| 621 |
.RE |
| 622 |
.RS 4 |
| 623 |
.RE |
| 624 |
.ie n .IP "$info = Storable::read_magic( $buffer )" 4 |
| 625 |
.el .IP "\f(CW$info\fR = Storable::read_magic( \f(CW$buffer\fR )" 4 |
| 626 |
.IX Item "$info = Storable::read_magic( $buffer )" |
| 627 |
.PD 0 |
| 628 |
.ie n .IP "$info = Storable::read_magic( $buffer, $must_be_file )" 4 |
| 629 |
.el .IP "\f(CW$info\fR = Storable::read_magic( \f(CW$buffer\fR, \f(CW$must_be_file\fR )" 4 |
| 630 |
.IX Item "$info = Storable::read_magic( $buffer, $must_be_file )" |
| 631 |
.PD |
| 632 |
The \f(CW$buffer\fR should be a Storable image or the first few bytes of it. |
| 633 |
If \f(CW$buffer\fR starts with a Storable header, then a hash describing the |
| 634 |
image is returned, otherwise \f(CW\*(C`undef\*(C'\fR is returned. |
| 635 |
.Sp |
| 636 |
The hash has the same structure as the one returned by |
| 637 |
\&\fBStorable::file_magic()\fR. The \f(CW\*(C`file\*(C'\fR element is true if the image is a |
| 638 |
file image. |
| 639 |
.Sp |
| 640 |
If the \f(CW$must_be_file\fR argument is provided and is TRUE, then return |
| 641 |
\&\f(CW\*(C`undef\*(C'\fR unless the image looks like it belongs to a file dump. |
| 642 |
.Sp |
| 643 |
The maximum size of a Storable header is currently 21 bytes. If the |
| 644 |
provided \f(CW$buffer\fR is only the first part of a Storable image it should |
| 645 |
at least be this long to ensure that \fBread_magic()\fR will recognize it as |
| 646 |
such. |
| 647 |
.SH EXAMPLES |
| 648 |
.IX Header "EXAMPLES" |
| 649 |
Here are some code samples showing a possible usage of Storable: |
| 650 |
.PP |
| 651 |
.Vb 1 |
| 652 |
\& use Storable qw(store retrieve freeze thaw dclone); |
| 653 |
\& |
| 654 |
\& %color = (\*(AqBlue\*(Aq => 0.1, \*(AqRed\*(Aq => 0.8, \*(AqBlack\*(Aq => 0, \*(AqWhite\*(Aq => 1); |
| 655 |
\& |
| 656 |
\& store(\e%color, \*(Aqmycolors\*(Aq) or die "Can\*(Aqt store %a in mycolors!\en"; |
| 657 |
\& |
| 658 |
\& $colref = retrieve(\*(Aqmycolors\*(Aq); |
| 659 |
\& die "Unable to retrieve from mycolors!\en" unless defined $colref; |
| 660 |
\& printf "Blue is still %lf\en", $colref\->{\*(AqBlue\*(Aq}; |
| 661 |
\& |
| 662 |
\& $colref2 = dclone(\e%color); |
| 663 |
\& |
| 664 |
\& $str = freeze(\e%color); |
| 665 |
\& printf "Serialization of %%color is %d bytes long.\en", length($str); |
| 666 |
\& $colref3 = thaw($str); |
| 667 |
.Ve |
| 668 |
.PP |
| 669 |
which prints (on my machine): |
| 670 |
.PP |
| 671 |
.Vb 2 |
| 672 |
\& Blue is still 0.100000 |
| 673 |
\& Serialization of %color is 102 bytes long. |
| 674 |
.Ve |
| 675 |
.PP |
| 676 |
Serialization of CODE references and deserialization in a safe |
| 677 |
compartment: |
| 678 |
.PP |
| 679 |
.Vb 11 |
| 680 |
\& use Storable qw(freeze thaw); |
| 681 |
\& use Safe; |
| 682 |
\& use strict; |
| 683 |
\& my $safe = new Safe; |
| 684 |
\& # because of opcodes used in "use strict": |
| 685 |
\& $safe\->permit(qw(:default require)); |
| 686 |
\& local $Storable::Deparse = 1; |
| 687 |
\& local $Storable::Eval = sub { $safe\->reval($_[0]) }; |
| 688 |
\& my $serialized = freeze(sub { 42 }); |
| 689 |
\& my $code = thaw($serialized); |
| 690 |
\& $code\->() == 42; |
| 691 |
.Ve |
| 692 |
.SH "SECURITY WARNING" |
| 693 |
.IX Header "SECURITY WARNING" |
| 694 |
\&\fBDo not accept Storable documents from untrusted sources!\fR There is |
| 695 |
\&\fBno\fR way to configure Storable so that it can be used safely to process |
| 696 |
untrusted data. While there \fIare\fR various options that can be used to |
| 697 |
mitigate specific security issues these options do \fInot\fR comprise a |
| 698 |
complete safety net for the user, and processing untrusted data may |
| 699 |
result in segmentation faults, remote code execution, or privilege |
| 700 |
escalation. The following lists some known features which represent |
| 701 |
security issues that should be considered by users of this module. |
| 702 |
.PP |
| 703 |
Most obviously, the optional (off by default) CODE reference |
| 704 |
serialization feature allows transfer of code to the deserializing |
| 705 |
process. Furthermore, any serialized object will cause Storable to |
| 706 |
helpfully load the module corresponding to the class of the object in |
| 707 |
the deserializing module. For manipulated module names, this can load |
| 708 |
almost arbitrary code. Finally, the deserialized object's destructors |
| 709 |
will be invoked when the objects get destroyed in the deserializing |
| 710 |
process. Maliciously crafted Storable documents may put such objects |
| 711 |
in the value of a hash key that is overridden by another key/value |
| 712 |
pair in the same hash, thus causing immediate destructor execution. |
| 713 |
.PP |
| 714 |
To disable blessing objects while thawing/retrieving remove the flag |
| 715 |
\&\f(CW\*(C`BLESS_OK\*(C'\fR = 2 from \f(CW$Storable::flags\fR or set the 2nd argument for |
| 716 |
thaw/retrieve to 0. |
| 717 |
.PP |
| 718 |
To disable tieing data while thawing/retrieving remove the flag \f(CW\*(C`TIE_OK\*(C'\fR |
| 719 |
= 4 from \f(CW$Storable::flags\fR or set the 2nd argument for thaw/retrieve |
| 720 |
to 0. |
| 721 |
.PP |
| 722 |
With the default setting of \f(CW$Storable::flags\fR = 6, creating or destroying |
| 723 |
random objects, even renamed objects can be controlled by an attacker. |
| 724 |
See CVE\-2015\-1592 and its metasploit module. |
| 725 |
.PP |
| 726 |
If your application requires accepting data from untrusted sources, you |
| 727 |
are best off with a less powerful and more-likely safe serialization |
| 728 |
format and implementation. If your data is sufficiently simple, |
| 729 |
Cpanel::JSON::XS or Data::MessagePack are fine alternatives. For |
| 730 |
more complex data structures containing various Perl specific data types |
| 731 |
like regular expressions or aliased data Sereal is the best |
| 732 |
alternative and offers maximum interoperability. Note that Sereal is |
| 733 |
unsafe by default, but you can configure |
| 734 |
the encoder and decoder to mitigate any security issues. |
| 735 |
.SH WARNING |
| 736 |
.IX Header "WARNING" |
| 737 |
If you're using references as keys within your hash tables, you're bound |
| 738 |
to be disappointed when retrieving your data. Indeed, Perl stringifies |
| 739 |
references used as hash table keys. If you later wish to access the |
| 740 |
items via another reference stringification (i.e. using the same |
| 741 |
reference that was used for the key originally to record the value into |
| 742 |
the hash table), it will work because both references stringify to the |
| 743 |
same string. |
| 744 |
.PP |
| 745 |
It won't work across a sequence of \f(CW\*(C`store\*(C'\fR and \f(CW\*(C`retrieve\*(C'\fR operations, |
| 746 |
however, because the addresses in the retrieved objects, which are |
| 747 |
part of the stringified references, will probably differ from the |
| 748 |
original addresses. The topology of your structure is preserved, |
| 749 |
but not hidden semantics like those. |
| 750 |
.PP |
| 751 |
On platforms where it matters, be sure to call \f(CWbinmode()\fR on the |
| 752 |
descriptors that you pass to Storable functions. |
| 753 |
.PP |
| 754 |
Storing data canonically that contains large hashes can be |
| 755 |
significantly slower than storing the same data normally, as |
| 756 |
temporary arrays to hold the keys for each hash have to be allocated, |
| 757 |
populated, sorted and freed. Some tests have shown a halving of the |
| 758 |
speed of storing \-\- the exact penalty will depend on the complexity of |
| 759 |
your data. There is no slowdown on retrieval. |
| 760 |
.SH "REGULAR EXPRESSIONS" |
| 761 |
.IX Header "REGULAR EXPRESSIONS" |
| 762 |
Storable now has experimental support for storing regular expressions, |
| 763 |
but there are significant limitations: |
| 764 |
.IP \(bu 4 |
| 765 |
perl 5.8 or later is required. |
| 766 |
.IP \(bu 4 |
| 767 |
regular expressions with code blocks, ie \f(CW\*(C`/(?{ ... })/\*(C'\fR or \f(CW\*(C`/(??{ |
| 768 |
\&... })/\*(C'\fR will throw an exception when thawed. |
| 769 |
.IP \(bu 4 |
| 770 |
regular expression syntax and flags have changed over the history of |
| 771 |
perl, so a regular expression that you freeze in one version of perl |
| 772 |
may fail to thaw or behave differently in another version of perl. |
| 773 |
.IP \(bu 4 |
| 774 |
depending on the version of perl, regular expressions can change in |
| 775 |
behaviour depending on the context, but later perls will bake that |
| 776 |
behaviour into the regexp. |
| 777 |
.PP |
| 778 |
Storable will throw an exception if a frozen regular expression cannot |
| 779 |
be thawed. |
| 780 |
.SH BUGS |
| 781 |
.IX Header "BUGS" |
| 782 |
You can't store GLOB, FORMLINE, etc.... If you can define semantics |
| 783 |
for those operations, feel free to enhance Storable so that it can |
| 784 |
deal with them. |
| 785 |
.PP |
| 786 |
The store functions will \f(CW\*(C`croak\*(C'\fR if they run into such references |
| 787 |
unless you set \f(CW$Storable::forgive_me\fR to some \f(CW\*(C`TRUE\*(C'\fR value. In that |
| 788 |
case, the fatal message is converted to a warning and some meaningless |
| 789 |
string is stored instead. |
| 790 |
.PP |
| 791 |
Setting \f(CW$Storable::canonical\fR may not yield frozen strings that |
| 792 |
compare equal due to possible stringification of numbers. When the |
| 793 |
string version of a scalar exists, it is the form stored; therefore, |
| 794 |
if you happen to use your numbers as strings between two freezing |
| 795 |
operations on the same data structures, you will get different |
| 796 |
results. |
| 797 |
.PP |
| 798 |
When storing doubles in network order, their value is stored as text. |
| 799 |
However, you should also not expect non-numeric floating-point values |
| 800 |
such as infinity and "not a number" to pass successfully through a |
| 801 |
\&\fBnstore()\fR/\fBretrieve()\fR pair. |
| 802 |
.PP |
| 803 |
As Storable neither knows nor cares about character sets (although it |
| 804 |
does know that characters may be more than eight bits wide), any difference |
| 805 |
in the interpretation of character codes between a host and a target |
| 806 |
system is your problem. In particular, if host and target use different |
| 807 |
code points to represent the characters used in the text representation |
| 808 |
of floating-point numbers, you will not be able be able to exchange |
| 809 |
floating-point data, even with \fBnstore()\fR. |
| 810 |
.PP |
| 811 |
\&\f(CW\*(C`Storable::drop_utf8\*(C'\fR is a blunt tool. There is no facility either to |
| 812 |
return \fBall\fR strings as utf8 sequences, or to attempt to convert utf8 |
| 813 |
data back to 8 bit and \f(CWcroak()\fR if the conversion fails. |
| 814 |
.PP |
| 815 |
Prior to Storable 2.01, no distinction was made between signed and |
| 816 |
unsigned integers on storing. By default Storable prefers to store a |
| 817 |
scalars string representation (if it has one) so this would only cause |
| 818 |
problems when storing large unsigned integers that had never been converted |
| 819 |
to string or floating point. In other words values that had been generated |
| 820 |
by integer operations such as logic ops and then not used in any string or |
| 821 |
arithmetic context before storing. |
| 822 |
.SS "64 bit data in perl 5.6.0 and 5.6.1" |
| 823 |
.IX Subsection "64 bit data in perl 5.6.0 and 5.6.1" |
| 824 |
This section only applies to you if you have existing data written out |
| 825 |
by Storable 2.02 or earlier on perl 5.6.0 or 5.6.1 on Unix or Linux which |
| 826 |
has been configured with 64 bit integer support (not the default) |
| 827 |
If you got a precompiled perl, rather than running Configure to build |
| 828 |
your own perl from source, then it almost certainly does not affect you, |
| 829 |
and you can stop reading now (unless you're curious). If you're using perl |
| 830 |
on Windows it does not affect you. |
| 831 |
.PP |
| 832 |
Storable writes a file header which contains the sizes of various C |
| 833 |
language types for the C compiler that built Storable (when not writing in |
| 834 |
network order), and will refuse to load files written by a Storable not |
| 835 |
on the same (or compatible) architecture. This check and a check on |
| 836 |
machine byteorder is needed because the size of various fields in the file |
| 837 |
are given by the sizes of the C language types, and so files written on |
| 838 |
different architectures are incompatible. This is done for increased speed. |
| 839 |
(When writing in network order, all fields are written out as standard |
| 840 |
lengths, which allows full interworking, but takes longer to read and write) |
| 841 |
.PP |
| 842 |
Perl 5.6.x introduced the ability to optional configure the perl interpreter |
| 843 |
to use C's \f(CW\*(C`long long\*(C'\fR type to allow scalars to store 64 bit integers on 32 |
| 844 |
bit systems. However, due to the way the Perl configuration system |
| 845 |
generated the C configuration files on non-Windows platforms, and the way |
| 846 |
Storable generates its header, nothing in the Storable file header reflected |
| 847 |
whether the perl writing was using 32 or 64 bit integers, despite the fact |
| 848 |
that Storable was storing some data differently in the file. Hence Storable |
| 849 |
running on perl with 64 bit integers will read the header from a file |
| 850 |
written by a 32 bit perl, not realise that the data is actually in a subtly |
| 851 |
incompatible format, and then go horribly wrong (possibly crashing) if it |
| 852 |
encountered a stored integer. This is a design failure. |
| 853 |
.PP |
| 854 |
Storable has now been changed to write out and read in a file header with |
| 855 |
information about the size of integers. It's impossible to detect whether |
| 856 |
an old file being read in was written with 32 or 64 bit integers (they have |
| 857 |
the same header) so it's impossible to automatically switch to a correct |
| 858 |
backwards compatibility mode. Hence this Storable defaults to the new, |
| 859 |
correct behaviour. |
| 860 |
.PP |
| 861 |
What this means is that if you have data written by Storable 1.x running |
| 862 |
on perl 5.6.0 or 5.6.1 configured with 64 bit integers on Unix or Linux |
| 863 |
then by default this Storable will refuse to read it, giving the error |
| 864 |
\&\fIByte order is not compatible\fR. If you have such data then you |
| 865 |
should set \f(CW$Storable::interwork_56_64bit\fR to a true value to make this |
| 866 |
Storable read and write files with the old header. You should also |
| 867 |
migrate your data, or any older perl you are communicating with, to this |
| 868 |
current version of Storable. |
| 869 |
.PP |
| 870 |
If you don't have data written with specific configuration of perl described |
| 871 |
above, then you do not and should not do anything. Don't set the flag \- |
| 872 |
not only will Storable on an identically configured perl refuse to load them, |
| 873 |
but Storable a differently configured perl will load them believing them |
| 874 |
to be correct for it, and then may well fail or crash part way through |
| 875 |
reading them. |
| 876 |
.SH CREDITS |
| 877 |
.IX Header "CREDITS" |
| 878 |
Thank you to (in chronological order): |
| 879 |
.PP |
| 880 |
.Vb 10 |
| 881 |
\& Jarkko Hietaniemi <[email protected]> |
| 882 |
\& Ulrich Pfeifer <[email protected]\-dortmund.de> |
| 883 |
\& Benjamin A. Holzman <[email protected]> |
| 884 |
\& Andrew Ford <A.Ford@ford\-mason.co.uk> |
| 885 |
\& Gisle Aas <[email protected]> |
| 886 |
\& Jeff Gresham <[email protected]> |
| 887 |
\& Murray Nesbitt <[email protected]> |
| 888 |
\& Marc Lehmann <[email protected]> |
| 889 |
\& Justin Banks <[email protected]> |
| 890 |
\& Jarkko Hietaniemi <[email protected]> (AGAIN, as perl 5.7.0 Pumpkin!) |
| 891 |
\& Salvador Ortiz Garcia <[email protected]> |
| 892 |
\& Dominic Dunlop <[email protected]> |
| 893 |
\& Erik Haugan <[email protected]> |
| 894 |
\& Benjamin A. Holzman <[email protected]> |
| 895 |
\& Reini Urban <[email protected]> |
| 896 |
\& Todd Rinaldo <[email protected]> |
| 897 |
\& Aaron Crane <[email protected]> |
| 898 |
.Ve |
| 899 |
.PP |
| 900 |
for their bug reports, suggestions and contributions. |
| 901 |
.PP |
| 902 |
Benjamin Holzman contributed the tied variable support, Andrew Ford |
| 903 |
contributed the canonical order for hashes, and Gisle Aas fixed |
| 904 |
a few misunderstandings of mine regarding the perl internals, |
| 905 |
and optimized the emission of "tags" in the output streams by |
| 906 |
simply counting the objects instead of tagging them (leading to |
| 907 |
a binary incompatibility for the Storable image starting at version |
| 908 |
0.6\-\-older images are, of course, still properly understood). |
| 909 |
Murray Nesbitt made Storable thread-safe. Marc Lehmann added overloading |
| 910 |
and references to tied items support. Benjamin Holzman added a performance |
| 911 |
improvement for overloaded classes; thanks to Grant Street Group for footing |
| 912 |
the bill. |
| 913 |
Reini Urban took over maintenance from p5p, and added security fixes |
| 914 |
and huge object support. |
| 915 |
.SH AUTHOR |
| 916 |
.IX Header "AUTHOR" |
| 917 |
Storable was written by Raphael Manfredi |
| 918 |
\&\fI<[email protected]>\fR |
| 919 |
Maintenance is now done by cperl <http://perl11.org/cperl> |
| 920 |
.PP |
| 921 |
Please e\-mail us with problems, bug fixes, comments and complaints, |
| 922 |
although if you have compliments you should send them to Raphael. |
| 923 |
Please don't e\-mail Raphael with problems, as he no longer works on |
| 924 |
Storable, and your message will be delayed while he forwards it to us. |
| 925 |
.SH "SEE ALSO" |
| 926 |
.IX Header "SEE ALSO" |
| 927 |
Clone. |