| 1 | .\" Automatically generated by Pod::Man v1.34, Pod::Parser v1.13 |
| 2 | .\" |
| 3 | .\" Standard preamble: |
| 4 | .\" ======================================================================== |
| 5 | .de Sh \" Subsection heading |
| 6 | .br |
| 7 | .if t .Sp |
| 8 | .ne 5 |
| 9 | .PP |
| 10 | \fB\\$1\fR |
| 11 | .PP |
| 12 | .. |
| 13 | .de Sp \" Vertical space (when we can't use .PP) |
| 14 | .if t .sp .5v |
| 15 | .if n .sp |
| 16 | .. |
| 17 | .de Vb \" Begin verbatim text |
| 18 | .ft CW |
| 19 | .nf |
| 20 | .ne \\$1 |
| 21 | .. |
| 22 | .de Ve \" End verbatim text |
| 23 | .ft R |
| 24 | .fi |
| 25 | .. |
| 26 | .\" Set up some character translations and predefined strings. \*(-- will |
| 27 | .\" give an unbreakable dash, \*(PI will give pi, \*(L" will give a left |
| 28 | .\" double quote, and \*(R" will give a right double quote. | will give a |
| 29 | .\" real vertical bar. \*(C+ will give a nicer C++. Capital omega is used to |
| 30 | .\" do unbreakable dashes and therefore won't be available. \*(C` and \*(C' |
| 31 | .\" expand to `' in nroff, nothing in troff, for use with C<>. |
| 32 | .tr \(*W-|\(bv\*(Tr |
| 33 | .ds C+ C\v'-.1v'\h'-1p'\s-2+\h'-1p'+\s0\v'.1v'\h'-1p' |
| 34 | .ie n \{\ |
| 35 | . ds -- \(*W- |
| 36 | . ds PI pi |
| 37 | . if (\n(.H=4u)&(1m=24u) .ds -- \(*W\h'-12u'\(*W\h'-12u'-\" diablo 10 pitch |
| 38 | . if (\n(.H=4u)&(1m=20u) .ds -- \(*W\h'-12u'\(*W\h'-8u'-\" diablo 12 pitch |
| 39 | . ds L" "" |
| 40 | . ds R" "" |
| 41 | . ds C` "" |
| 42 | . ds C' "" |
| 43 | 'br\} |
| 44 | .el\{\ |
| 45 | . ds -- \|\(em\| |
| 46 | . ds PI \(*p |
| 47 | . ds L" `` |
| 48 | . ds R" '' |
| 49 | 'br\} |
| 50 | .\" |
| 51 | .\" If the F register is turned on, we'll generate index entries on stderr for |
| 52 | .\" titles (.TH), headers (.SH), subsections (.Sh), items (.Ip), and index |
| 53 | .\" entries marked with X<> in POD. Of course, you'll have to process the |
| 54 | .\" output yourself in some meaningful fashion. |
| 55 | .if \nF \{\ |
| 56 | . de IX |
| 57 | . tm Index:\\$1\t\\n%\t"\\$2" |
| 58 | .. |
| 59 | . nr % 0 |
| 60 | . rr F |
| 61 | .\} |
| 62 | .\" |
| 63 | .\" For nroff, turn off justification. Always turn off hyphenation; it makes |
| 64 | .\" way too many mistakes in technical documents. |
| 65 | .hy 0 |
| 66 | .if n .na |
| 67 | .\" |
| 68 | .\" Accent mark definitions (@(#)ms.acc 1.5 88/02/08 SMI; from UCB 4.2). |
| 69 | .\" Fear. Run. Save yourself. No user-serviceable parts. |
| 70 | . \" fudge factors for nroff and troff |
| 71 | .if n \{\ |
| 72 | . ds #H 0 |
| 73 | . ds #V .8m |
| 74 | . ds #F .3m |
| 75 | . ds #[ \f1 |
| 76 | . ds #] \fP |
| 77 | .\} |
| 78 | .if t \{\ |
| 79 | . ds #H ((1u-(\\\\n(.fu%2u))*.13m) |
| 80 | . ds #V .6m |
| 81 | . ds #F 0 |
| 82 | . ds #[ \& |
| 83 | . ds #] \& |
| 84 | .\} |
| 85 | . \" simple accents for nroff and troff |
| 86 | .if n \{\ |
| 87 | . ds ' \& |
| 88 | . ds ` \& |
| 89 | . ds ^ \& |
| 90 | . ds , \& |
| 91 | . ds ~ ~ |
| 92 | . ds / |
| 93 | .\} |
| 94 | .if t \{\ |
| 95 | . ds ' \\k:\h'-(\\n(.wu*8/10-\*(#H)'\'\h"|\\n:u" |
| 96 | . ds ` \\k:\h'-(\\n(.wu*8/10-\*(#H)'\`\h'|\\n:u' |
| 97 | . ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'^\h'|\\n:u' |
| 98 | . ds , \\k:\h'-(\\n(.wu*8/10)',\h'|\\n:u' |
| 99 | . ds ~ \\k:\h'-(\\n(.wu-\*(#H-.1m)'~\h'|\\n:u' |
| 100 | . ds / \\k:\h'-(\\n(.wu*8/10-\*(#H)'\z\(sl\h'|\\n:u' |
| 101 | .\} |
| 102 | . \" troff and (daisy-wheel) nroff accents |
| 103 | .ds : \\k:\h'-(\\n(.wu*8/10-\*(#H+.1m+\*(#F)'\v'-\*(#V'\z.\h'.2m+\*(#F'.\h'|\\n:u'\v'\*(#V' |
| 104 | .ds 8 \h'\*(#H'\(*b\h'-\*(#H' |
| 105 | .ds o \\k:\h'-(\\n(.wu+\w'\(de'u-\*(#H)/2u'\v'-.3n'\*(#[\z\(de\v'.3n'\h'|\\n:u'\*(#] |
| 106 | .ds d- \h'\*(#H'\(pd\h'-\w'~'u'\v'-.25m'\f2\(hy\fP\v'.25m'\h'-\*(#H' |
| 107 | .ds D- D\\k:\h'-\w'D'u'\v'-.11m'\z\(hy\v'.11m'\h'|\\n:u' |
| 108 | .ds th \*(#[\v'.3m'\s+1I\s-1\v'-.3m'\h'-(\w'I'u*2/3)'\s-1o\s+1\*(#] |
| 109 | .ds Th \*(#[\s+2I\s-2\h'-\w'I'u*3/5'\v'-.3m'o\v'.3m'\*(#] |
| 110 | .ds ae a\h'-(\w'a'u*4/10)'e |
| 111 | .ds Ae A\h'-(\w'A'u*4/10)'E |
| 112 | . \" corrections for vroff |
| 113 | .if v .ds ~ \\k:\h'-(\\n(.wu*9/10-\*(#H)'\s-2\u~\d\s+2\h'|\\n:u' |
| 114 | .if v .ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'\v'-.4m'^\v'.4m'\h'|\\n:u' |
| 115 | . \" for low resolution devices (crt and lpr) |
| 116 | .if \n(.H>23 .if \n(.V>19 \ |
| 117 | \{\ |
| 118 | . ds : e |
| 119 | . ds 8 ss |
| 120 | . ds o a |
| 121 | . ds d- d\h'-1'\(ga |
| 122 | . ds D- D\h'-1'\(hy |
| 123 | . ds th \o'bp' |
| 124 | . ds Th \o'LP' |
| 125 | . ds ae ae |
| 126 | . ds Ae AE |
| 127 | .\} |
| 128 | .rm #[ #] #H #V #F C |
| 129 | .\" ======================================================================== |
| 130 | .\" |
| 131 | .IX Title "PERLTOOC 1" |
| 132 | .TH PERLTOOC 1 "2002-06-08" "perl v5.8.0" "Perl Programmers Reference Guide" |
| 133 | .SH "NAME" |
| 134 | perltooc \- Tom's OO Tutorial for Class Data in Perl |
| 135 | .SH "DESCRIPTION" |
| 136 | .IX Header "DESCRIPTION" |
| 137 | When designing an object class, you are sometimes faced with the situation |
| 138 | of wanting common state shared by all objects of that class. |
| 139 | Such \fIclass attributes\fR act somewhat like global variables for the entire |
| 140 | class, but unlike program-wide globals, class attributes have meaning only to |
| 141 | the class itself. |
| 142 | .PP |
| 143 | Here are a few examples where class attributes might come in handy: |
| 144 | .IP "\(bu" 4 |
| 145 | to keep a count of the objects you've created, or how many are |
| 146 | still extant. |
| 147 | .IP "\(bu" 4 |
| 148 | to extract the name or file descriptor for a logfile used by a debugging |
| 149 | method. |
| 150 | .IP "\(bu" 4 |
| 151 | to access collective data, like the total amount of cash dispensed by |
| 152 | all ATMs in a network in a given day. |
| 153 | .IP "\(bu" 4 |
| 154 | to access the last object created by a class, or the most accessed object, |
| 155 | or to retrieve a list of all objects. |
| 156 | .PP |
| 157 | Unlike a true global, class attributes should not be accessed directly. |
| 158 | Instead, their state should be inspected, and perhaps altered, only |
| 159 | through the mediated access of \fIclass methods\fR. These class attributes |
| 160 | accessor methods are similar in spirit and function to accessors used |
| 161 | to manipulate the state of instance attributes on an object. They provide a |
| 162 | clear firewall between interface and implementation. |
| 163 | .PP |
| 164 | You should allow access to class attributes through either the class |
| 165 | name or any object of that class. If we assume that \f(CW$an_object\fR is of |
| 166 | type Some_Class, and the &Some_Class::population_count method accesses |
| 167 | class attributes, then these two invocations should both be possible, |
| 168 | and almost certainly equivalent. |
| 169 | .PP |
| 170 | .Vb 2 |
| 171 | \& Some_Class->population_count() |
| 172 | \& $an_object->population_count() |
| 173 | .Ve |
| 174 | .PP |
| 175 | The question is, where do you store the state which that method accesses? |
| 176 | Unlike more restrictive languages like \*(C+, where these are called |
| 177 | static data members, Perl provides no syntactic mechanism to declare |
| 178 | class attributes, any more than it provides a syntactic mechanism to |
| 179 | declare instance attributes. Perl provides the developer with a broad |
| 180 | set of powerful but flexible features that can be uniquely crafted to |
| 181 | the particular demands of the situation. |
| 182 | .PP |
| 183 | A class in Perl is typically implemented in a module. A module consists |
| 184 | of two complementary feature sets: a package for interfacing with the |
| 185 | outside world, and a lexical file scope for privacy. Either of these |
| 186 | two mechanisms can be used to implement class attributes. That means you |
| 187 | get to decide whether to put your class attributes in package variables |
| 188 | or to put them in lexical variables. |
| 189 | .PP |
| 190 | And those aren't the only decisions to make. If you choose to use package |
| 191 | variables, you can make your class attribute accessor methods either ignorant |
| 192 | of inheritance or sensitive to it. If you choose lexical variables, |
| 193 | you can elect to permit access to them from anywhere in the entire file |
| 194 | scope, or you can limit direct data access exclusively to the methods |
| 195 | implementing those attributes. |
| 196 | .SH "Class Data in a Can" |
| 197 | .IX Header "Class Data in a Can" |
| 198 | One of the easiest ways to solve a hard problem is to let someone else |
| 199 | do it for you! In this case, Class::Data::Inheritable (available on a |
| 200 | \&\s-1CPAN\s0 near you) offers a canned solution to the class data problem |
| 201 | using closures. So before you wade into this document, consider |
| 202 | having a look at that module. |
| 203 | .SH "Class Data as Package Variables" |
| 204 | .IX Header "Class Data as Package Variables" |
| 205 | Because a class in Perl is really just a package, using package variables |
| 206 | to hold class attributes is the most natural choice. This makes it simple |
| 207 | for each class to have its own class attributes. Let's say you have a class |
| 208 | called Some_Class that needs a couple of different attributes that you'd |
| 209 | like to be global to the entire class. The simplest thing to do is to |
| 210 | use package variables like \f(CW$Some_Class::CData1\fR and \f(CW$Some_Class::CData2\fR |
| 211 | to hold these attributes. But we certainly don't want to encourage |
| 212 | outsiders to touch those data directly, so we provide methods |
| 213 | to mediate access. |
| 214 | .PP |
| 215 | In the accessor methods below, we'll for now just ignore the first |
| 216 | argument\*(--that part to the left of the arrow on method invocation, which |
| 217 | is either a class name or an object reference. |
| 218 | .PP |
| 219 | .Vb 11 |
| 220 | \& package Some_Class; |
| 221 | \& sub CData1 { |
| 222 | \& shift; # XXX: ignore calling class/object |
| 223 | \& $Some_Class::CData1 = shift if @_; |
| 224 | \& return $Some_Class::CData1; |
| 225 | \& } |
| 226 | \& sub CData2 { |
| 227 | \& shift; # XXX: ignore calling class/object |
| 228 | \& $Some_Class::CData2 = shift if @_; |
| 229 | \& return $Some_Class::CData2; |
| 230 | \& } |
| 231 | .Ve |
| 232 | .PP |
| 233 | This technique is highly legible and should be completely straightforward |
| 234 | to even the novice Perl programmer. By fully qualifying the package |
| 235 | variables, they stand out clearly when reading the code. Unfortunately, |
| 236 | if you misspell one of these, you've introduced an error that's hard |
| 237 | to catch. It's also somewhat disconcerting to see the class name itself |
| 238 | hard-coded in so many places. |
| 239 | .PP |
| 240 | Both these problems can be easily fixed. Just add the \f(CW\*(C`use strict\*(C'\fR |
| 241 | pragma, then pre-declare your package variables. (The \f(CW\*(C`our\*(C'\fR operator |
| 242 | will be new in 5.6, and will work for package globals just like \f(CW\*(C`my\*(C'\fR |
| 243 | works for scoped lexicals.) |
| 244 | .PP |
| 245 | .Vb 13 |
| 246 | \& package Some_Class; |
| 247 | \& use strict; |
| 248 | \& our($CData1, $CData2); # our() is new to perl5.6 |
| 249 | \& sub CData1 { |
| 250 | \& shift; # XXX: ignore calling class/object |
| 251 | \& $CData1 = shift if @_; |
| 252 | \& return $CData1; |
| 253 | \& } |
| 254 | \& sub CData2 { |
| 255 | \& shift; # XXX: ignore calling class/object |
| 256 | \& $CData2 = shift if @_; |
| 257 | \& return $CData2; |
| 258 | \& } |
| 259 | .Ve |
| 260 | .PP |
| 261 | As with any other global variable, some programmers prefer to start their |
| 262 | package variables with capital letters. This helps clarity somewhat, but |
| 263 | by no longer fully qualifying the package variables, their significance |
| 264 | can be lost when reading the code. You can fix this easily enough by |
| 265 | choosing better names than were used here. |
| 266 | .Sh "Putting All Your Eggs in One Basket" |
| 267 | .IX Subsection "Putting All Your Eggs in One Basket" |
| 268 | Just as the mindless enumeration of accessor methods for instance attributes |
| 269 | grows tedious after the first few (see perltoot), so too does the |
| 270 | repetition begin to grate when listing out accessor methods for class |
| 271 | data. Repetition runs counter to the primary virtue of a programmer: |
| 272 | Laziness, here manifesting as that innate urge every programmer feels |
| 273 | to factor out duplicate code whenever possible. |
| 274 | .PP |
| 275 | Here's what to do. First, make just one hash to hold all class attributes. |
| 276 | .PP |
| 277 | .Vb 6 |
| 278 | \& package Some_Class; |
| 279 | \& use strict; |
| 280 | \& our %ClassData = ( # our() is new to perl5.6 |
| 281 | \& CData1 => "", |
| 282 | \& CData2 => "", |
| 283 | \& ); |
| 284 | .Ve |
| 285 | .PP |
| 286 | Using closures (see perlref) and direct access to the package symbol |
| 287 | table (see perlmod), now clone an accessor method for each key in |
| 288 | the \f(CW%ClassData\fR hash. Each of these methods is used to fetch or store |
| 289 | values to the specific, named class attribute. |
| 290 | .PP |
| 291 | .Vb 8 |
| 292 | \& for my $datum (keys %ClassData) { |
| 293 | \& no strict "refs"; # to register new methods in package |
| 294 | \& *$datum = sub { |
| 295 | \& shift; # XXX: ignore calling class/object |
| 296 | \& $ClassData{$datum} = shift if @_; |
| 297 | \& return $ClassData{$datum}; |
| 298 | \& } |
| 299 | \& } |
| 300 | .Ve |
| 301 | .PP |
| 302 | It's true that you could work out a solution employing an &AUTOLOAD |
| 303 | method, but this approach is unlikely to prove satisfactory. Your |
| 304 | function would have to distinguish between class attributes and object |
| 305 | attributes; it could interfere with inheritance; and it would have to |
| 306 | careful about \s-1DESTROY\s0. Such complexity is uncalled for in most cases, |
| 307 | and certainly in this one. |
| 308 | .PP |
| 309 | You may wonder why we're rescinding strict refs for the loop. We're |
| 310 | manipulating the package's symbol table to introduce new function names |
| 311 | using symbolic references (indirect naming), which the strict pragma |
| 312 | would otherwise forbid. Normally, symbolic references are a dodgy |
| 313 | notion at best. This isn't just because they can be used accidentally |
| 314 | when you aren't meaning to. It's also because for most uses |
| 315 | to which beginning Perl programmers attempt to put symbolic references, |
| 316 | we have much better approaches, like nested hashes or hashes of arrays. |
| 317 | But there's nothing wrong with using symbolic references to manipulate |
| 318 | something that is meaningful only from the perspective of the package |
| 319 | symbol table, like method names or package variables. In other |
| 320 | words, when you want to refer to the symbol table, use symbol references. |
| 321 | .PP |
| 322 | Clustering all the class attributes in one place has several advantages. |
| 323 | They're easy to spot, initialize, and change. The aggregation also |
| 324 | makes them convenient to access externally, such as from a debugger |
| 325 | or a persistence package. The only possible problem is that we don't |
| 326 | automatically know the name of each class's class object, should it have |
| 327 | one. This issue is addressed below in \*(L"The Eponymous Meta\-Object\*(R". |
| 328 | .Sh "Inheritance Concerns" |
| 329 | .IX Subsection "Inheritance Concerns" |
| 330 | Suppose you have an instance of a derived class, and you access class |
| 331 | data using an inherited method call. Should that end up referring |
| 332 | to the base class's attributes, or to those in the derived class? |
| 333 | How would it work in the earlier examples? The derived class inherits |
| 334 | all the base class's methods, including those that access class attributes. |
| 335 | But what package are the class attributes stored in? |
| 336 | .PP |
| 337 | The answer is that, as written, class attributes are stored in the package into |
| 338 | which those methods were compiled. When you invoke the &CData1 method |
| 339 | on the name of the derived class or on one of that class's objects, the |
| 340 | version shown above is still run, so you'll access \f(CW$Some_Class::CData1\fR\-\-or |
| 341 | in the method cloning version, \f(CW$Some_Class::ClassData{CData1}\fR. |
| 342 | .PP |
| 343 | Think of these class methods as executing in the context of their base |
| 344 | class, not in that of their derived class. Sometimes this is exactly |
| 345 | what you want. If Feline subclasses Carnivore, then the population of |
| 346 | Carnivores in the world should go up when a new Feline is born. |
| 347 | But what if you wanted to figure out how many Felines you have apart |
| 348 | from Carnivores? The current approach doesn't support that. |
| 349 | .PP |
| 350 | You'll have to decide on a case-by-case basis whether it makes any sense |
| 351 | for class attributes to be package\-relative. If you want it to be so, |
| 352 | then stop ignoring the first argument to the function. Either it will |
| 353 | be a package name if the method was invoked directly on a class name, |
| 354 | or else it will be an object reference if the method was invoked on an |
| 355 | object reference. In the latter case, the \fIref()\fR function provides the |
| 356 | class of that object. |
| 357 | .PP |
| 358 | .Vb 9 |
| 359 | \& package Some_Class; |
| 360 | \& sub CData1 { |
| 361 | \& my $obclass = shift; |
| 362 | \& my $class = ref($obclass) || $obclass; |
| 363 | \& my $varname = $class . "::CData1"; |
| 364 | \& no strict "refs"; # to access package data symbolically |
| 365 | \& $$varname = shift if @_; |
| 366 | \& return $$varname; |
| 367 | \& } |
| 368 | .Ve |
| 369 | .PP |
| 370 | And then do likewise for all other class attributes (such as CData2, |
| 371 | etc.) that you wish to access as package variables in the invoking package |
| 372 | instead of the compiling package as we had previously. |
| 373 | .PP |
| 374 | Once again we temporarily disable the strict references ban, because |
| 375 | otherwise we couldn't use the fully-qualified symbolic name for |
| 376 | the package global. This is perfectly reasonable: since all package |
| 377 | variables by definition live in a package, there's nothing wrong with |
| 378 | accessing them via that package's symbol table. That's what it's there |
| 379 | for (well, somewhat). |
| 380 | .PP |
| 381 | What about just using a single hash for everything and then cloning |
| 382 | methods? What would that look like? The only difference would be the |
| 383 | closure used to produce new method entries for the class's symbol table. |
| 384 | .PP |
| 385 | .Vb 8 |
| 386 | \& no strict "refs"; |
| 387 | \& *$datum = sub { |
| 388 | \& my $obclass = shift; |
| 389 | \& my $class = ref($obclass) || $obclass; |
| 390 | \& my $varname = $class . "::ClassData"; |
| 391 | \& $varname->{$datum} = shift if @_; |
| 392 | \& return $varname->{$datum}; |
| 393 | \& } |
| 394 | .Ve |
| 395 | .Sh "The Eponymous Meta-Object" |
| 396 | .IX Subsection "The Eponymous Meta-Object" |
| 397 | It could be argued that the \f(CW%ClassData\fR hash in the previous example is |
| 398 | neither the most imaginative nor the most intuitive of names. Is there |
| 399 | something else that might make more sense, be more useful, or both? |
| 400 | .PP |
| 401 | As it happens, yes, there is. For the \*(L"class meta\-object\*(R", we'll use |
| 402 | a package variable of the same name as the package itself. Within the |
| 403 | scope of a package Some_Class declaration, we'll use the eponymously |
| 404 | named hash \f(CW%Some_Class\fR as that class's meta\-object. (Using an eponymously |
| 405 | named hash is somewhat reminiscent of classes that name their constructors |
| 406 | eponymously in the Python or \*(C+ fashion. That is, class Some_Class would |
| 407 | use &Some_Class::Some_Class as a constructor, probably even exporting that |
| 408 | name as well. The StrNum class in Recipe 13.14 in \fIThe Perl Cookbook\fR |
| 409 | does this, if you're looking for an example.) |
| 410 | .PP |
| 411 | This predictable approach has many benefits, including having a well-known |
| 412 | identifier to aid in debugging, transparent persistence, |
| 413 | or checkpointing. It's also the obvious name for monadic classes and |
| 414 | translucent attributes, discussed later. |
| 415 | .PP |
| 416 | Here's an example of such a class. Notice how the name of the |
| 417 | hash storing the meta-object is the same as the name of the package |
| 418 | used to implement the class. |
| 419 | .PP |
| 420 | .Vb 2 |
| 421 | \& package Some_Class; |
| 422 | \& use strict; |
| 423 | .Ve |
| 424 | .PP |
| 425 | .Vb 5 |
| 426 | \& # create class meta-object using that most perfect of names |
| 427 | \& our %Some_Class = ( # our() is new to perl5.6 |
| 428 | \& CData1 => "", |
| 429 | \& CData2 => "", |
| 430 | \& ); |
| 431 | .Ve |
| 432 | .PP |
| 433 | .Vb 8 |
| 434 | \& # this accessor is calling-package-relative |
| 435 | \& sub CData1 { |
| 436 | \& my $obclass = shift; |
| 437 | \& my $class = ref($obclass) || $obclass; |
| 438 | \& no strict "refs"; # to access eponymous meta-object |
| 439 | \& $class->{CData1} = shift if @_; |
| 440 | \& return $class->{CData1}; |
| 441 | \& } |
| 442 | .Ve |
| 443 | .PP |
| 444 | .Vb 7 |
| 445 | \& # but this accessor is not |
| 446 | \& sub CData2 { |
| 447 | \& shift; # XXX: ignore calling class/object |
| 448 | \& no strict "refs"; # to access eponymous meta-object |
| 449 | \& __PACKAGE__ -> {CData2} = shift if @_; |
| 450 | \& return __PACKAGE__ -> {CData2}; |
| 451 | \& } |
| 452 | .Ve |
| 453 | .PP |
| 454 | In the second accessor method, the _\|_PACKAGE_\|_ notation was used for |
| 455 | two reasons. First, to avoid hardcoding the literal package name |
| 456 | in the code in case we later want to change that name. Second, to |
| 457 | clarify to the reader that what matters here is the package currently |
| 458 | being compiled into, not the package of the invoking object or class. |
| 459 | If the long sequence of non-alphabetic characters bothers you, you can |
| 460 | always put the _\|_PACKAGE_\|_ in a variable first. |
| 461 | .PP |
| 462 | .Vb 7 |
| 463 | \& sub CData2 { |
| 464 | \& shift; # XXX: ignore calling class/object |
| 465 | \& no strict "refs"; # to access eponymous meta-object |
| 466 | \& my $class = __PACKAGE__; |
| 467 | \& $class->{CData2} = shift if @_; |
| 468 | \& return $class->{CData2}; |
| 469 | \& } |
| 470 | .Ve |
| 471 | .PP |
| 472 | Even though we're using symbolic references for good not evil, some |
| 473 | folks tend to become unnerved when they see so many places with strict |
| 474 | ref checking disabled. Given a symbolic reference, you can always |
| 475 | produce a real reference (the reverse is not true, though). So we'll |
| 476 | create a subroutine that does this conversion for us. If invoked as a |
| 477 | function of no arguments, it returns a reference to the compiling class's |
| 478 | eponymous hash. Invoked as a class method, it returns a reference to |
| 479 | the eponymous hash of its caller. And when invoked as an object method, |
| 480 | this function returns a reference to the eponymous hash for whatever |
| 481 | class the object belongs to. |
| 482 | .PP |
| 483 | .Vb 2 |
| 484 | \& package Some_Class; |
| 485 | \& use strict; |
| 486 | .Ve |
| 487 | .PP |
| 488 | .Vb 4 |
| 489 | \& our %Some_Class = ( # our() is new to perl5.6 |
| 490 | \& CData1 => "", |
| 491 | \& CData2 => "", |
| 492 | \& ); |
| 493 | .Ve |
| 494 | .PP |
| 495 | .Vb 7 |
| 496 | \& # tri-natured: function, class method, or object method |
| 497 | \& sub _classobj { |
| 498 | \& my $obclass = shift || __PACKAGE__; |
| 499 | \& my $class = ref($obclass) || $obclass; |
| 500 | \& no strict "refs"; # to convert sym ref to real one |
| 501 | \& return \e%$class; |
| 502 | \& } |
| 503 | .Ve |
| 504 | .PP |
| 505 | .Vb 11 |
| 506 | \& for my $datum (keys %{ _classobj() } ) { |
| 507 | \& # turn off strict refs so that we can |
| 508 | \& # register a method in the symbol table |
| 509 | \& no strict "refs"; |
| 510 | \& *$datum = sub { |
| 511 | \& use strict "refs"; |
| 512 | \& my $self = shift->_classobj(); |
| 513 | \& $self->{$datum} = shift if @_; |
| 514 | \& return $self->{$datum}; |
| 515 | \& } |
| 516 | \& } |
| 517 | .Ve |
| 518 | .Sh "Indirect References to Class Data" |
| 519 | .IX Subsection "Indirect References to Class Data" |
| 520 | A reasonably common strategy for handling class attributes is to store |
| 521 | a reference to each package variable on the object itself. This is |
| 522 | a strategy you've probably seen before, such as in perltoot and |
| 523 | perlbot, but there may be variations in the example below that you |
| 524 | haven't thought of before. |
| 525 | .PP |
| 526 | .Vb 2 |
| 527 | \& package Some_Class; |
| 528 | \& our($CData1, $CData2); # our() is new to perl5.6 |
| 529 | .Ve |
| 530 | .PP |
| 531 | .Vb 9 |
| 532 | \& sub new { |
| 533 | \& my $obclass = shift; |
| 534 | \& return bless my $self = { |
| 535 | \& ObData1 => "", |
| 536 | \& ObData2 => "", |
| 537 | \& CData1 => \e$CData1, |
| 538 | \& CData2 => \e$CData2, |
| 539 | \& } => (ref $obclass || $obclass); |
| 540 | \& } |
| 541 | .Ve |
| 542 | .PP |
| 543 | .Vb 5 |
| 544 | \& sub ObData1 { |
| 545 | \& my $self = shift; |
| 546 | \& $self->{ObData1} = shift if @_; |
| 547 | \& return $self->{ObData1}; |
| 548 | \& } |
| 549 | .Ve |
| 550 | .PP |
| 551 | .Vb 5 |
| 552 | \& sub ObData2 { |
| 553 | \& my $self = shift; |
| 554 | \& $self->{ObData2} = shift if @_; |
| 555 | \& return $self->{ObData2}; |
| 556 | \& } |
| 557 | .Ve |
| 558 | .PP |
| 559 | .Vb 8 |
| 560 | \& sub CData1 { |
| 561 | \& my $self = shift; |
| 562 | \& my $dataref = ref $self |
| 563 | \& ? $self->{CData1} |
| 564 | \& : \e$CData1; |
| 565 | \& $$dataref = shift if @_; |
| 566 | \& return $$dataref; |
| 567 | \& } |
| 568 | .Ve |
| 569 | .PP |
| 570 | .Vb 8 |
| 571 | \& sub CData2 { |
| 572 | \& my $self = shift; |
| 573 | \& my $dataref = ref $self |
| 574 | \& ? $self->{CData2} |
| 575 | \& : \e$CData2; |
| 576 | \& $$dataref = shift if @_; |
| 577 | \& return $$dataref; |
| 578 | \& } |
| 579 | .Ve |
| 580 | .PP |
| 581 | As written above, a derived class will inherit these methods, which |
| 582 | will consequently access package variables in the base class's package. |
| 583 | This is not necessarily expected behavior in all circumstances. Here's an |
| 584 | example that uses a variable meta\-object, taking care to access the |
| 585 | proper package's data. |
| 586 | .PP |
| 587 | .Vb 2 |
| 588 | \& package Some_Class; |
| 589 | \& use strict; |
| 590 | .Ve |
| 591 | .PP |
| 592 | .Vb 4 |
| 593 | \& our %Some_Class = ( # our() is new to perl5.6 |
| 594 | \& CData1 => "", |
| 595 | \& CData2 => "", |
| 596 | \& ); |
| 597 | .Ve |
| 598 | .PP |
| 599 | .Vb 7 |
| 600 | \& sub _classobj { |
| 601 | \& my $self = shift; |
| 602 | \& my $class = ref($self) || $self; |
| 603 | \& no strict "refs"; |
| 604 | \& # get (hard) ref to eponymous meta-object |
| 605 | \& return \e%$class; |
| 606 | \& } |
| 607 | .Ve |
| 608 | .PP |
| 609 | .Vb 11 |
| 610 | \& sub new { |
| 611 | \& my $obclass = shift; |
| 612 | \& my $classobj = $obclass->_classobj(); |
| 613 | \& bless my $self = { |
| 614 | \& ObData1 => "", |
| 615 | \& ObData2 => "", |
| 616 | \& CData1 => \e$classobj->{CData1}, |
| 617 | \& CData2 => \e$classobj->{CData2}, |
| 618 | \& } => (ref $obclass || $obclass); |
| 619 | \& return $self; |
| 620 | \& } |
| 621 | .Ve |
| 622 | .PP |
| 623 | .Vb 5 |
| 624 | \& sub ObData1 { |
| 625 | \& my $self = shift; |
| 626 | \& $self->{ObData1} = shift if @_; |
| 627 | \& return $self->{ObData1}; |
| 628 | \& } |
| 629 | .Ve |
| 630 | .PP |
| 631 | .Vb 5 |
| 632 | \& sub ObData2 { |
| 633 | \& my $self = shift; |
| 634 | \& $self->{ObData2} = shift if @_; |
| 635 | \& return $self->{ObData2}; |
| 636 | \& } |
| 637 | .Ve |
| 638 | .PP |
| 639 | .Vb 7 |
| 640 | \& sub CData1 { |
| 641 | \& my $self = shift; |
| 642 | \& $self = $self->_classobj() unless ref $self; |
| 643 | \& my $dataref = $self->{CData1}; |
| 644 | \& $$dataref = shift if @_; |
| 645 | \& return $$dataref; |
| 646 | \& } |
| 647 | .Ve |
| 648 | .PP |
| 649 | .Vb 7 |
| 650 | \& sub CData2 { |
| 651 | \& my $self = shift; |
| 652 | \& $self = $self->_classobj() unless ref $self; |
| 653 | \& my $dataref = $self->{CData2}; |
| 654 | \& $$dataref = shift if @_; |
| 655 | \& return $$dataref; |
| 656 | \& } |
| 657 | .Ve |
| 658 | .PP |
| 659 | Not only are we now strict refs clean, using an eponymous meta-object |
| 660 | seems to make the code cleaner. Unlike the previous version, this one |
| 661 | does something interesting in the face of inheritance: it accesses the |
| 662 | class meta-object in the invoking class instead of the one into which |
| 663 | the method was initially compiled. |
| 664 | .PP |
| 665 | You can easily access data in the class meta\-object, making |
| 666 | it easy to dump the complete class state using an external mechanism such |
| 667 | as when debugging or implementing a persistent class. This works because |
| 668 | the class meta-object is a package variable, has a well-known name, and |
| 669 | clusters all its data together. (Transparent persistence |
| 670 | is not always feasible, but it's certainly an appealing idea.) |
| 671 | .PP |
| 672 | There's still no check that object accessor methods have not been |
| 673 | invoked on a class name. If strict ref checking is enabled, you'd |
| 674 | blow up. If not, then you get the eponymous meta\-object. What you do |
| 675 | with\*(--or about\*(--this is up to you. The next two sections demonstrate |
| 676 | innovative uses for this powerful feature. |
| 677 | .Sh "Monadic Classes" |
| 678 | .IX Subsection "Monadic Classes" |
| 679 | Some of the standard modules shipped with Perl provide class interfaces |
| 680 | without any attribute methods whatsoever. The most commonly used module |
| 681 | not numbered amongst the pragmata, the Exporter module, is a class with |
| 682 | neither constructors nor attributes. Its job is simply to provide a |
| 683 | standard interface for modules wishing to export part of their namespace |
| 684 | into that of their caller. Modules use the Exporter's &import method by |
| 685 | setting their inheritance list in their package's \f(CW@ISA\fR array to mention |
| 686 | \&\*(L"Exporter\*(R". But class Exporter provides no constructor, so you can't |
| 687 | have several instances of the class. In fact, you can't have any\*(--it |
| 688 | just doesn't make any sense. All you get is its methods. Its interface |
| 689 | contains no statefulness, so state data is wholly superfluous. |
| 690 | .PP |
| 691 | Another sort of class that pops up from time to time is one that supports |
| 692 | a unique instance. Such classes are called \fImonadic classes\fR, or less |
| 693 | formally, \fIsingletons\fR or \fIhighlander classes\fR. |
| 694 | .PP |
| 695 | If a class is monadic, where do you store its state, that is, |
| 696 | its attributes? How do you make sure that there's never more than |
| 697 | one instance? While you could merely use a slew of package variables, |
| 698 | it's a lot cleaner to use the eponymously named hash. Here's a complete |
| 699 | example of a monadic class: |
| 700 | .PP |
| 701 | .Vb 2 |
| 702 | \& package Cosmos; |
| 703 | \& %Cosmos = (); |
| 704 | .Ve |
| 705 | .PP |
| 706 | .Vb 6 |
| 707 | \& # accessor method for "name" attribute |
| 708 | \& sub name { |
| 709 | \& my $self = shift; |
| 710 | \& $self->{name} = shift if @_; |
| 711 | \& return $self->{name}; |
| 712 | \& } |
| 713 | .Ve |
| 714 | .PP |
| 715 | .Vb 6 |
| 716 | \& # read-only accessor method for "birthday" attribute |
| 717 | \& sub birthday { |
| 718 | \& my $self = shift; |
| 719 | \& die "can't reset birthday" if @_; # XXX: croak() is better |
| 720 | \& return $self->{birthday}; |
| 721 | \& } |
| 722 | .Ve |
| 723 | .PP |
| 724 | .Vb 6 |
| 725 | \& # accessor method for "stars" attribute |
| 726 | \& sub stars { |
| 727 | \& my $self = shift; |
| 728 | \& $self->{stars} = shift if @_; |
| 729 | \& return $self->{stars}; |
| 730 | \& } |
| 731 | .Ve |
| 732 | .PP |
| 733 | .Vb 6 |
| 734 | \& # oh my - one of our stars just went out! |
| 735 | \& sub supernova { |
| 736 | \& my $self = shift; |
| 737 | \& my $count = $self->stars(); |
| 738 | \& $self->stars($count - 1) if $count > 0; |
| 739 | \& } |
| 740 | .Ve |
| 741 | .PP |
| 742 | .Vb 10 |
| 743 | \& # constructor/initializer method - fix by reboot |
| 744 | \& sub bigbang { |
| 745 | \& my $self = shift; |
| 746 | \& %$self = ( |
| 747 | \& name => "the world according to tchrist", |
| 748 | \& birthday => time(), |
| 749 | \& stars => 0, |
| 750 | \& ); |
| 751 | \& return $self; # yes, it's probably a class. SURPRISE! |
| 752 | \& } |
| 753 | .Ve |
| 754 | .PP |
| 755 | .Vb 3 |
| 756 | \& # After the class is compiled, but before any use or require |
| 757 | \& # returns, we start off the universe with a bang. |
| 758 | \& __PACKAGE__ -> bigbang(); |
| 759 | .Ve |
| 760 | .PP |
| 761 | Hold on, that doesn't look like anything special. Those attribute |
| 762 | accessors look no different than they would if this were a regular class |
| 763 | instead of a monadic one. The crux of the matter is there's nothing |
| 764 | that says that \f(CW$self\fR must hold a reference to a blessed object. It merely |
| 765 | has to be something you can invoke methods on. Here the package name |
| 766 | itself, Cosmos, works as an object. Look at the &supernova method. Is that |
| 767 | a class method or an object method? The answer is that static analysis |
| 768 | cannot reveal the answer. Perl doesn't care, and neither should you. |
| 769 | In the three attribute methods, \f(CW%$self\fR is really accessing the \f(CW%Cosmos\fR |
| 770 | package variable. |
| 771 | .PP |
| 772 | If like Stephen Hawking, you posit the existence of multiple, sequential, |
| 773 | and unrelated universes, then you can invoke the &bigbang method yourself |
| 774 | at any time to start everything all over again. You might think of |
| 775 | &bigbang as more of an initializer than a constructor, since the function |
| 776 | doesn't allocate new memory; it only initializes what's already there. |
| 777 | But like any other constructor, it does return a scalar value to use |
| 778 | for later method invocations. |
| 779 | .PP |
| 780 | Imagine that some day in the future, you decide that one universe just |
| 781 | isn't enough. You could write a new class from scratch, but you already |
| 782 | have an existing class that does what you want\*(--except that it's monadic, |
| 783 | and you want more than just one cosmos. |
| 784 | .PP |
| 785 | That's what code reuse via subclassing is all about. Look how short |
| 786 | the new code is: |
| 787 | .PP |
| 788 | .Vb 3 |
| 789 | \& package Multiverse; |
| 790 | \& use Cosmos; |
| 791 | \& @ISA = qw(Cosmos); |
| 792 | .Ve |
| 793 | .PP |
| 794 | .Vb 7 |
| 795 | \& sub new { |
| 796 | \& my $protoverse = shift; |
| 797 | \& my $class = ref($protoverse) || $protoverse; |
| 798 | \& my $self = {}; |
| 799 | \& return bless($self, $class)->bigbang(); |
| 800 | \& } |
| 801 | \& 1; |
| 802 | .Ve |
| 803 | .PP |
| 804 | Because we were careful to be good little creators when we designed our |
| 805 | Cosmos class, we can now reuse it without touching a single line of code |
| 806 | when it comes time to write our Multiverse class. The same code that |
| 807 | worked when invoked as a class method continues to work perfectly well |
| 808 | when invoked against separate instances of a derived class. |
| 809 | .PP |
| 810 | The astonishing thing about the Cosmos class above is that the value |
| 811 | returned by the &bigbang \*(L"constructor\*(R" is not a reference to a blessed |
| 812 | object at all. It's just the class's own name. A class name is, for |
| 813 | virtually all intents and purposes, a perfectly acceptable object. |
| 814 | It has state, behavior, and identify, the three crucial components |
| 815 | of an object system. It even manifests inheritance, polymorphism, |
| 816 | and encapsulation. And what more can you ask of an object? |
| 817 | .PP |
| 818 | To understand object orientation in Perl, it's important to recognize the |
| 819 | unification of what other programming languages might think of as class |
| 820 | methods and object methods into just plain methods. \*(L"Class methods\*(R" |
| 821 | and \*(L"object methods\*(R" are distinct only in the compartmentalizing mind |
| 822 | of the Perl programmer, not in the Perl language itself. |
| 823 | .PP |
| 824 | Along those same lines, a constructor is nothing special either, which |
| 825 | is one reason why Perl has no pre-ordained name for them. \*(L"Constructor\*(R" |
| 826 | is just an informal term loosely used to describe a method that returns |
| 827 | a scalar value that you can make further method calls against. So long |
| 828 | as it's either a class name or an object reference, that's good enough. |
| 829 | It doesn't even have to be a reference to a brand new object. |
| 830 | .PP |
| 831 | You can have as many\*(--or as few\*(--constructors as you want, and you can |
| 832 | name them whatever you care to. Blindly and obediently using \fInew()\fR |
| 833 | for each and every constructor you ever write is to speak Perl with |
| 834 | such a severe \*(C+ accent that you do a disservice to both languages. |
| 835 | There's no reason to insist that each class have but one constructor, |
| 836 | or that a constructor be named \fInew()\fR, or that a constructor be |
| 837 | used solely as a class method and not an object method. |
| 838 | .PP |
| 839 | The next section shows how useful it can be to further distance ourselves |
| 840 | from any formal distinction between class method calls and object method |
| 841 | calls, both in constructors and in accessor methods. |
| 842 | .Sh "Translucent Attributes" |
| 843 | .IX Subsection "Translucent Attributes" |
| 844 | A package's eponymous hash can be used for more than just containing |
| 845 | per\-class, global state data. It can also serve as a sort of template |
| 846 | containing default settings for object attributes. These default |
| 847 | settings can then be used in constructors for initialization of a |
| 848 | particular object. The class's eponymous hash can also be used to |
| 849 | implement \fItranslucent attributes\fR. A translucent attribute is one |
| 850 | that has a class-wide default. Each object can set its own value for the |
| 851 | attribute, in which case \f(CW\*(C`$object\->attribute()\*(C'\fR returns that value. |
| 852 | But if no value has been set, then \f(CW\*(C`$object\->attribute()\*(C'\fR returns |
| 853 | the class-wide default. |
| 854 | .PP |
| 855 | We'll apply something of a copy-on-write approach to these translucent |
| 856 | attributes. If you're just fetching values from them, you get |
| 857 | translucency. But if you store a new value to them, that new value is |
| 858 | set on the current object. On the other hand, if you use the class as |
| 859 | an object and store the attribute value directly on the class, then the |
| 860 | meta\-object's value changes, and later fetch operations on objects with |
| 861 | uninitialized values for those attributes will retrieve the meta\-object's |
| 862 | new values. Objects with their own initialized values, however, won't |
| 863 | see any change. |
| 864 | .PP |
| 865 | Let's look at some concrete examples of using these properties before we |
| 866 | show how to implement them. Suppose that a class named Some_Class |
| 867 | had a translucent data attribute called \*(L"color\*(R". First you set the color |
| 868 | in the meta\-object, then you create three objects using a constructor |
| 869 | that happens to be named &spawn. |
| 870 | .PP |
| 871 | .Vb 2 |
| 872 | \& use Vermin; |
| 873 | \& Vermin->color("vermilion"); |
| 874 | .Ve |
| 875 | .PP |
| 876 | .Vb 3 |
| 877 | \& $ob1 = Vermin->spawn(); # so that's where Jedi come from |
| 878 | \& $ob2 = Vermin->spawn(); |
| 879 | \& $ob3 = Vermin->spawn(); |
| 880 | .Ve |
| 881 | .PP |
| 882 | .Vb 1 |
| 883 | \& print $obj3->color(); # prints "vermilion" |
| 884 | .Ve |
| 885 | .PP |
| 886 | Each of these objects' colors is now \*(L"vermilion\*(R", because that's the |
| 887 | meta\-object's value that attribute, and these objects do not have |
| 888 | individual color values set. |
| 889 | .PP |
| 890 | Changing the attribute on one object has no effect on other objects |
| 891 | previously created. |
| 892 | .PP |
| 893 | .Vb 3 |
| 894 | \& $ob3->color("chartreuse"); |
| 895 | \& print $ob3->color(); # prints "chartreuse" |
| 896 | \& print $ob1->color(); # prints "vermilion", translucently |
| 897 | .Ve |
| 898 | .PP |
| 899 | If you now use \f(CW$ob3\fR to spawn off another object, the new object will |
| 900 | take the color its parent held, which now happens to be \*(L"chartreuse\*(R". |
| 901 | That's because the constructor uses the invoking object as its template |
| 902 | for initializing attributes. When that invoking object is the |
| 903 | class name, the object used as a template is the eponymous meta\-object. |
| 904 | When the invoking object is a reference to an instantiated object, the |
| 905 | &spawn constructor uses that existing object as a template. |
| 906 | .PP |
| 907 | .Vb 2 |
| 908 | \& $ob4 = $ob3->spawn(); # $ob3 now template, not %Vermin |
| 909 | \& print $ob4->color(); # prints "chartreuse" |
| 910 | .Ve |
| 911 | .PP |
| 912 | Any actual values set on the template object will be copied to the |
| 913 | new object. But attributes undefined in the template object, being |
| 914 | translucent, will remain undefined and consequently translucent in the |
| 915 | new one as well. |
| 916 | .PP |
| 917 | Now let's change the color attribute on the entire class: |
| 918 | .PP |
| 919 | .Vb 5 |
| 920 | \& Vermin->color("azure"); |
| 921 | \& print $ob1->color(); # prints "azure" |
| 922 | \& print $ob2->color(); # prints "azure" |
| 923 | \& print $ob3->color(); # prints "chartreuse" |
| 924 | \& print $ob4->color(); # prints "chartreuse" |
| 925 | .Ve |
| 926 | .PP |
| 927 | That color change took effect only in the first pair of objects, which |
| 928 | were still translucently accessing the meta\-object's values. The second |
| 929 | pair had per-object initialized colors, and so didn't change. |
| 930 | .PP |
| 931 | One important question remains. Changes to the meta-object are reflected |
| 932 | in translucent attributes in the entire class, but what about |
| 933 | changes to discrete objects? If you change the color of \f(CW$ob3\fR, does the |
| 934 | value of \f(CW$ob4\fR see that change? Or vice\-versa. If you change the color |
| 935 | of \f(CW$ob4\fR, does then the value of \f(CW$ob3\fR shift? |
| 936 | .PP |
| 937 | .Vb 3 |
| 938 | \& $ob3->color("amethyst"); |
| 939 | \& print $ob3->color(); # prints "amethyst" |
| 940 | \& print $ob4->color(); # hmm: "chartreuse" or "amethyst"? |
| 941 | .Ve |
| 942 | .PP |
| 943 | While one could argue that in certain rare cases it should, let's not |
| 944 | do that. Good taste aside, we want the answer to the question posed in |
| 945 | the comment above to be \*(L"chartreuse\*(R", not \*(L"amethyst\*(R". So we'll treat |
| 946 | these attributes similar to the way process attributes like environment |
| 947 | variables, user and group IDs, or the current working directory are |
| 948 | treated across a \fIfork()\fR. You can change only yourself, but you will see |
| 949 | those changes reflected in your unspawned children. Changes to one object |
| 950 | will propagate neither up to the parent nor down to any existing child objects. |
| 951 | Those objects made later, however, will see the changes. |
| 952 | .PP |
| 953 | If you have an object with an actual attribute value, and you want to |
| 954 | make that object's attribute value translucent again, what do you do? |
| 955 | Let's design the class so that when you invoke an accessor method with |
| 956 | \&\f(CW\*(C`undef\*(C'\fR as its argument, that attribute returns to translucency. |
| 957 | .PP |
| 958 | .Vb 1 |
| 959 | \& $ob4->color(undef); # back to "azure" |
| 960 | .Ve |
| 961 | .PP |
| 962 | Here's a complete implementation of Vermin as described above. |
| 963 | .PP |
| 964 | .Vb 1 |
| 965 | \& package Vermin; |
| 966 | .Ve |
| 967 | .PP |
| 968 | .Vb 4 |
| 969 | \& # here's the class meta-object, eponymously named. |
| 970 | \& # it holds all class attributes, and also all instance attributes |
| 971 | \& # so the latter can be used for both initialization |
| 972 | \& # and translucency. |
| 973 | .Ve |
| 974 | .PP |
| 975 | .Vb 4 |
| 976 | \& our %Vermin = ( # our() is new to perl5.6 |
| 977 | \& PopCount => 0, # capital for class attributes |
| 978 | \& color => "beige", # small for instance attributes |
| 979 | \& ); |
| 980 | .Ve |
| 981 | .PP |
| 982 | .Vb 13 |
| 983 | \& # constructor method |
| 984 | \& # invoked as class method or object method |
| 985 | \& sub spawn { |
| 986 | \& my $obclass = shift; |
| 987 | \& my $class = ref($obclass) || $obclass; |
| 988 | \& my $self = {}; |
| 989 | \& bless($self, $class); |
| 990 | \& $class->{PopCount}++; |
| 991 | \& # init fields from invoking object, or omit if |
| 992 | \& # invoking object is the class to provide translucency |
| 993 | \& %$self = %$obclass if ref $obclass; |
| 994 | \& return $self; |
| 995 | \& } |
| 996 | .Ve |
| 997 | .PP |
| 998 | .Vb 5 |
| 999 | \& # translucent accessor for "color" attribute |
| 1000 | \& # invoked as class method or object method |
| 1001 | \& sub color { |
| 1002 | \& my $self = shift; |
| 1003 | \& my $class = ref($self) || $self; |
| 1004 | .Ve |
| 1005 | .PP |
| 1006 | .Vb 5 |
| 1007 | \& # handle class invocation |
| 1008 | \& unless (ref $self) { |
| 1009 | \& $class->{color} = shift if @_; |
| 1010 | \& return $class->{color} |
| 1011 | \& } |
| 1012 | .Ve |
| 1013 | .PP |
| 1014 | .Vb 8 |
| 1015 | \& # handle object invocation |
| 1016 | \& $self->{color} = shift if @_; |
| 1017 | \& if (defined $self->{color}) { # not exists! |
| 1018 | \& return $self->{color}; |
| 1019 | \& } else { |
| 1020 | \& return $class->{color}; |
| 1021 | \& } |
| 1022 | \& } |
| 1023 | .Ve |
| 1024 | .PP |
| 1025 | .Vb 8 |
| 1026 | \& # accessor for "PopCount" class attribute |
| 1027 | \& # invoked as class method or object method |
| 1028 | \& # but uses object solely to locate meta-object |
| 1029 | \& sub population { |
| 1030 | \& my $obclass = shift; |
| 1031 | \& my $class = ref($obclass) || $obclass; |
| 1032 | \& return $class->{PopCount}; |
| 1033 | \& } |
| 1034 | .Ve |
| 1035 | .PP |
| 1036 | .Vb 7 |
| 1037 | \& # instance destructor |
| 1038 | \& # invoked only as object method |
| 1039 | \& sub DESTROY { |
| 1040 | \& my $self = shift; |
| 1041 | \& my $class = ref $self; |
| 1042 | \& $class->{PopCount}--; |
| 1043 | \& } |
| 1044 | .Ve |
| 1045 | .PP |
| 1046 | Here are a couple of helper methods that might be convenient. They aren't |
| 1047 | accessor methods at all. They're used to detect accessibility of data |
| 1048 | attributes. The &is_translucent method determines whether a particular |
| 1049 | object attribute is coming from the meta\-object. The &has_attribute |
| 1050 | method detects whether a class implements a particular property at all. |
| 1051 | It could also be used to distinguish undefined properties from non-existent |
| 1052 | ones. |
| 1053 | .PP |
| 1054 | .Vb 6 |
| 1055 | \& # detect whether an object attribute is translucent |
| 1056 | \& # (typically?) invoked only as object method |
| 1057 | \& sub is_translucent { |
| 1058 | \& my($self, $attr) = @_; |
| 1059 | \& return !defined $self->{$attr}; |
| 1060 | \& } |
| 1061 | .Ve |
| 1062 | .PP |
| 1063 | .Vb 7 |
| 1064 | \& # test for presence of attribute in class |
| 1065 | \& # invoked as class method or object method |
| 1066 | \& sub has_attribute { |
| 1067 | \& my($self, $attr) = @_; |
| 1068 | \& my $class = ref $self if $self; |
| 1069 | \& return exists $class->{$attr}; |
| 1070 | \& } |
| 1071 | .Ve |
| 1072 | .PP |
| 1073 | If you prefer to install your accessors more generically, you can make |
| 1074 | use of the upper-case versus lower-case convention to register into the |
| 1075 | package appropriate methods cloned from generic closures. |
| 1076 | .PP |
| 1077 | .Vb 20 |
| 1078 | \& for my $datum (keys %{ +__PACKAGE__ }) { |
| 1079 | \& *$datum = ($datum =~ /^[A-Z]/) |
| 1080 | \& ? sub { # install class accessor |
| 1081 | \& my $obclass = shift; |
| 1082 | \& my $class = ref($obclass) || $obclass; |
| 1083 | \& return $class->{$datum}; |
| 1084 | \& } |
| 1085 | \& : sub { # install translucent accessor |
| 1086 | \& my $self = shift; |
| 1087 | \& my $class = ref($self) || $self; |
| 1088 | \& unless (ref $self) { |
| 1089 | \& $class->{$datum} = shift if @_; |
| 1090 | \& return $class->{$datum} |
| 1091 | \& } |
| 1092 | \& $self->{$datum} = shift if @_; |
| 1093 | \& return defined $self->{$datum} |
| 1094 | \& ? $self -> {$datum} |
| 1095 | \& : $class -> {$datum} |
| 1096 | \& } |
| 1097 | \& } |
| 1098 | .Ve |
| 1099 | .PP |
| 1100 | Translations of this closure-based approach into \*(C+, Java, and Python |
| 1101 | have been left as exercises for the reader. Be sure to send us mail as |
| 1102 | soon as you're done. |
| 1103 | .SH "Class Data as Lexical Variables" |
| 1104 | .IX Header "Class Data as Lexical Variables" |
| 1105 | .Sh "Privacy and Responsibility" |
| 1106 | .IX Subsection "Privacy and Responsibility" |
| 1107 | Unlike conventions used by some Perl programmers, in the previous |
| 1108 | examples, we didn't prefix the package variables used for class attributes |
| 1109 | with an underscore, nor did we do so for the names of the hash keys used |
| 1110 | for instance attributes. You don't need little markers on data names to |
| 1111 | suggest nominal privacy on attribute variables or hash keys, because these |
| 1112 | are \fBalready\fR notionally private! Outsiders have no business whatsoever |
| 1113 | playing with anything within a class save through the mediated access of |
| 1114 | its documented interface; in other words, through method invocations. |
| 1115 | And not even through just any method, either. Methods that begin with |
| 1116 | an underscore are traditionally considered off-limits outside the class. |
| 1117 | If outsiders skip the documented method interface to poke around the |
| 1118 | internals of your class and end up breaking something, that's not your |
| 1119 | fault\*(--it's theirs. |
| 1120 | .PP |
| 1121 | Perl believes in individual responsibility rather than mandated control. |
| 1122 | Perl respects you enough to let you choose your own preferred level of |
| 1123 | pain, or of pleasure. Perl believes that you are creative, intelligent, |
| 1124 | and capable of making your own decisions\*(--and fully expects you to |
| 1125 | take complete responsibility for your own actions. In a perfect world, |
| 1126 | these admonitions alone would suffice, and everyone would be intelligent, |
| 1127 | responsible, happy, and creative. And careful. One probably shouldn't |
| 1128 | forget careful, and that's a good bit harder to expect. Even Einstein |
| 1129 | would take wrong turns by accident and end up lost in the wrong part |
| 1130 | of town. |
| 1131 | .PP |
| 1132 | Some folks get the heebie-jeebies when they see package variables |
| 1133 | hanging out there for anyone to reach over and alter them. Some folks |
| 1134 | live in constant fear that someone somewhere might do something wicked. |
| 1135 | The solution to that problem is simply to fire the wicked, of course. |
| 1136 | But unfortunately, it's not as simple as all that. These cautious |
| 1137 | types are also afraid that they or others will do something not so |
| 1138 | much wicked as careless, whether by accident or out of desperation. |
| 1139 | If we fire everyone who ever gets careless, pretty soon there won't be |
| 1140 | anybody left to get any work done. |
| 1141 | .PP |
| 1142 | Whether it's needless paranoia or sensible caution, this uneasiness can |
| 1143 | be a problem for some people. We can take the edge off their discomfort |
| 1144 | by providing the option of storing class attributes as lexical variables |
| 1145 | instead of as package variables. The \fImy()\fR operator is the source of |
| 1146 | all privacy in Perl, and it is a powerful form of privacy indeed. |
| 1147 | .PP |
| 1148 | It is widely perceived, and indeed has often been written, that Perl |
| 1149 | provides no data hiding, that it affords the class designer no privacy |
| 1150 | nor isolation, merely a rag-tag assortment of weak and unenforcible |
| 1151 | social conventions instead. This perception is demonstrably false and |
| 1152 | easily disproven. In the next section, we show how to implement forms |
| 1153 | of privacy that are far stronger than those provided in nearly any |
| 1154 | other object-oriented language. |
| 1155 | .Sh "File-Scoped Lexicals" |
| 1156 | .IX Subsection "File-Scoped Lexicals" |
| 1157 | A lexical variable is visible only through the end of its static scope. |
| 1158 | That means that the only code able to access that variable is code |
| 1159 | residing textually below the \fImy()\fR operator through the end of its block |
| 1160 | if it has one, or through the end of the current file if it doesn't. |
| 1161 | .PP |
| 1162 | Starting again with our simplest example given at the start of this |
| 1163 | document, we replace \fIour()\fR variables with \fImy()\fR versions. |
| 1164 | .PP |
| 1165 | .Vb 12 |
| 1166 | \& package Some_Class; |
| 1167 | \& my($CData1, $CData2); # file scope, not in any package |
| 1168 | \& sub CData1 { |
| 1169 | \& shift; # XXX: ignore calling class/object |
| 1170 | \& $CData1 = shift if @_; |
| 1171 | \& return $CData1; |
| 1172 | \& } |
| 1173 | \& sub CData2 { |
| 1174 | \& shift; # XXX: ignore calling class/object |
| 1175 | \& $CData2 = shift if @_; |
| 1176 | \& return $CData2; |
| 1177 | \& } |
| 1178 | .Ve |
| 1179 | .PP |
| 1180 | So much for that old \f(CW$Some_Class::CData1\fR package variable and its brethren! |
| 1181 | Those are gone now, replaced with lexicals. No one outside the |
| 1182 | scope can reach in and alter the class state without resorting to the |
| 1183 | documented interface. Not even subclasses or superclasses of |
| 1184 | this one have unmediated access to \f(CW$CData1\fR. They have to invoke the &CData1 |
| 1185 | method against Some_Class or an instance thereof, just like anybody else. |
| 1186 | .PP |
| 1187 | To be scrupulously honest, that last statement assumes you haven't packed |
| 1188 | several classes together into the same file scope, nor strewn your class |
| 1189 | implementation across several different files. Accessibility of those |
| 1190 | variables is based uniquely on the static file scope. It has nothing to |
| 1191 | do with the package. That means that code in a different file but |
| 1192 | the same package (class) could not access those variables, yet code in the |
| 1193 | same file but a different package (class) could. There are sound reasons |
| 1194 | why we usually suggest a one-to-one mapping between files and packages |
| 1195 | and modules and classes. You don't have to stick to this suggestion if |
| 1196 | you really know what you're doing, but you're apt to confuse yourself |
| 1197 | otherwise, especially at first. |
| 1198 | .PP |
| 1199 | If you'd like to aggregate your class attributes into one lexically scoped, |
| 1200 | composite structure, you're perfectly free to do so. |
| 1201 | .PP |
| 1202 | .Vb 15 |
| 1203 | \& package Some_Class; |
| 1204 | \& my %ClassData = ( |
| 1205 | \& CData1 => "", |
| 1206 | \& CData2 => "", |
| 1207 | \& ); |
| 1208 | \& sub CData1 { |
| 1209 | \& shift; # XXX: ignore calling class/object |
| 1210 | \& $ClassData{CData1} = shift if @_; |
| 1211 | \& return $ClassData{CData1}; |
| 1212 | \& } |
| 1213 | \& sub CData2 { |
| 1214 | \& shift; # XXX: ignore calling class/object |
| 1215 | \& $ClassData{CData2} = shift if @_; |
| 1216 | \& return $ClassData{CData2}; |
| 1217 | \& } |
| 1218 | .Ve |
| 1219 | .PP |
| 1220 | To make this more scalable as other class attributes are added, we can |
| 1221 | again register closures into the package symbol table to create accessor |
| 1222 | methods for them. |
| 1223 | .PP |
| 1224 | .Vb 13 |
| 1225 | \& package Some_Class; |
| 1226 | \& my %ClassData = ( |
| 1227 | \& CData1 => "", |
| 1228 | \& CData2 => "", |
| 1229 | \& ); |
| 1230 | \& for my $datum (keys %ClassData) { |
| 1231 | \& no strict "refs"; |
| 1232 | \& *$datum = sub { |
| 1233 | \& shift; # XXX: ignore calling class/object |
| 1234 | \& $ClassData{$datum} = shift if @_; |
| 1235 | \& return $ClassData{$datum}; |
| 1236 | \& }; |
| 1237 | \& } |
| 1238 | .Ve |
| 1239 | .PP |
| 1240 | Requiring even your own class to use accessor methods like anybody else is |
| 1241 | probably a good thing. But demanding and expecting that everyone else, |
| 1242 | be they subclass or superclass, friend or foe, will all come to your |
| 1243 | object through mediation is more than just a good idea. It's absolutely |
| 1244 | critical to the model. Let there be in your mind no such thing as |
| 1245 | \&\*(L"public\*(R" data, nor even \*(L"protected\*(R" data, which is a seductive but |
| 1246 | ultimately destructive notion. Both will come back to bite at you. |
| 1247 | That's because as soon as you take that first step out of the solid |
| 1248 | position in which all state is considered completely private, save from the |
| 1249 | perspective of its own accessor methods, you have violated the envelope. |
| 1250 | And, having pierced that encapsulating envelope, you shall doubtless |
| 1251 | someday pay the price when future changes in the implementation break |
| 1252 | unrelated code. Considering that avoiding this infelicitous outcome was |
| 1253 | precisely why you consented to suffer the slings and arrows of obsequious |
| 1254 | abstraction by turning to object orientation in the first place, such |
| 1255 | breakage seems unfortunate in the extreme. |
| 1256 | .Sh "More Inheritance Concerns" |
| 1257 | .IX Subsection "More Inheritance Concerns" |
| 1258 | Suppose that Some_Class were used as a base class from which to derive |
| 1259 | Another_Class. If you invoke a &CData method on the derived class or |
| 1260 | on an object of that class, what do you get? Would the derived class |
| 1261 | have its own state, or would it piggyback on its base class's versions |
| 1262 | of the class attributes? |
| 1263 | .PP |
| 1264 | The answer is that under the scheme outlined above, the derived class |
| 1265 | would \fBnot\fR have its own state data. As before, whether you consider |
| 1266 | this a good thing or a bad one depends on the semantics of the classes |
| 1267 | involved. |
| 1268 | .PP |
| 1269 | The cleanest, sanest, simplest way to address per-class state in a |
| 1270 | lexical is for the derived class to override its base class's version |
| 1271 | of the method that accesses the class attributes. Since the actual method |
| 1272 | called is the one in the object's derived class if this exists, you |
| 1273 | automatically get per-class state this way. Any urge to provide an |
| 1274 | unadvertised method to sneak out a reference to the \f(CW%ClassData\fR hash |
| 1275 | should be strenuously resisted. |
| 1276 | .PP |
| 1277 | As with any other overridden method, the implementation in the |
| 1278 | derived class always has the option of invoking its base class's |
| 1279 | version of the method in addition to its own. Here's an example: |
| 1280 | .PP |
| 1281 | .Vb 2 |
| 1282 | \& package Another_Class; |
| 1283 | \& @ISA = qw(Some_Class); |
| 1284 | .Ve |
| 1285 | .PP |
| 1286 | .Vb 3 |
| 1287 | \& my %ClassData = ( |
| 1288 | \& CData1 => "", |
| 1289 | \& ); |
| 1290 | .Ve |
| 1291 | .PP |
| 1292 | .Vb 5 |
| 1293 | \& sub CData1 { |
| 1294 | \& my($self, $newvalue) = @_; |
| 1295 | \& if (@_ > 1) { |
| 1296 | \& # set locally first |
| 1297 | \& $ClassData{CData1} = $newvalue; |
| 1298 | .Ve |
| 1299 | .PP |
| 1300 | .Vb 8 |
| 1301 | \& # then pass the buck up to the first |
| 1302 | \& # overridden version, if there is one |
| 1303 | \& if ($self->can("SUPER::CData1")) { |
| 1304 | \& $self->SUPER::CData1($newvalue); |
| 1305 | \& } |
| 1306 | \& } |
| 1307 | \& return $ClassData{CData1}; |
| 1308 | \& } |
| 1309 | .Ve |
| 1310 | .PP |
| 1311 | Those dabbling in multiple inheritance might be concerned |
| 1312 | about there being more than one override. |
| 1313 | .PP |
| 1314 | .Vb 6 |
| 1315 | \& for my $parent (@ISA) { |
| 1316 | \& my $methname = $parent . "::CData1"; |
| 1317 | \& if ($self->can($methname)) { |
| 1318 | \& $self->$methname($newvalue); |
| 1319 | \& } |
| 1320 | \& } |
| 1321 | .Ve |
| 1322 | .PP |
| 1323 | Because the &UNIVERSAL::can method returns a reference |
| 1324 | to the function directly, you can use this directly |
| 1325 | for a significant performance improvement: |
| 1326 | .PP |
| 1327 | .Vb 5 |
| 1328 | \& for my $parent (@ISA) { |
| 1329 | \& if (my $coderef = $self->can($parent . "::CData1")) { |
| 1330 | \& $self->$coderef($newvalue); |
| 1331 | \& } |
| 1332 | \& } |
| 1333 | .Ve |
| 1334 | .Sh "Locking the Door and Throwing Away the Key" |
| 1335 | .IX Subsection "Locking the Door and Throwing Away the Key" |
| 1336 | As currently implemented, any code within the same scope as the |
| 1337 | file-scoped lexical \f(CW%ClassData\fR can alter that hash directly. Is that |
| 1338 | ok? Is it acceptable or even desirable to allow other parts of the |
| 1339 | implementation of this class to access class attributes directly? |
| 1340 | .PP |
| 1341 | That depends on how careful you want to be. Think back to the Cosmos |
| 1342 | class. If the &supernova method had directly altered \f(CW$Cosmos::Stars\fR or |
| 1343 | \&\f(CW$Cosmos::Cosmos{stars}\fR, then we wouldn't have been able to reuse the |
| 1344 | class when it came to inventing a Multiverse. So letting even the class |
| 1345 | itself access its own class attributes without the mediating intervention of |
| 1346 | properly designed accessor methods is probably not a good idea after all. |
| 1347 | .PP |
| 1348 | Restricting access to class attributes from the class itself is usually |
| 1349 | not enforcible even in strongly object-oriented languages. But in Perl, |
| 1350 | you can. |
| 1351 | .PP |
| 1352 | Here's one way: |
| 1353 | .PP |
| 1354 | .Vb 1 |
| 1355 | \& package Some_Class; |
| 1356 | .Ve |
| 1357 | .PP |
| 1358 | .Vb 8 |
| 1359 | \& { # scope for hiding $CData1 |
| 1360 | \& my $CData1; |
| 1361 | \& sub CData1 { |
| 1362 | \& shift; # XXX: unused |
| 1363 | \& $CData1 = shift if @_; |
| 1364 | \& return $CData1; |
| 1365 | \& } |
| 1366 | \& } |
| 1367 | .Ve |
| 1368 | .PP |
| 1369 | .Vb 8 |
| 1370 | \& { # scope for hiding $CData2 |
| 1371 | \& my $CData2; |
| 1372 | \& sub CData2 { |
| 1373 | \& shift; # XXX: unused |
| 1374 | \& $CData2 = shift if @_; |
| 1375 | \& return $CData2; |
| 1376 | \& } |
| 1377 | \& } |
| 1378 | .Ve |
| 1379 | .PP |
| 1380 | No one\*(--absolutely no one\*(--is allowed to read or write the class |
| 1381 | attributes without the mediation of the managing accessor method, since |
| 1382 | only that method has access to the lexical variable it's managing. |
| 1383 | This use of mediated access to class attributes is a form of privacy far |
| 1384 | stronger than most \s-1OO\s0 languages provide. |
| 1385 | .PP |
| 1386 | The repetition of code used to create per-datum accessor methods chafes |
| 1387 | at our Laziness, so we'll again use closures to create similar |
| 1388 | methods. |
| 1389 | .PP |
| 1390 | .Vb 1 |
| 1391 | \& package Some_Class; |
| 1392 | .Ve |
| 1393 | .PP |
| 1394 | .Vb 5 |
| 1395 | \& { # scope for ultra-private meta-object for class attributes |
| 1396 | \& my %ClassData = ( |
| 1397 | \& CData1 => "", |
| 1398 | \& CData2 => "", |
| 1399 | \& ); |
| 1400 | .Ve |
| 1401 | .PP |
| 1402 | .Vb 9 |
| 1403 | \& for my $datum (keys %ClassData ) { |
| 1404 | \& no strict "refs"; |
| 1405 | \& *$datum = sub { |
| 1406 | \& use strict "refs"; |
| 1407 | \& my ($self, $newvalue) = @_; |
| 1408 | \& $ClassData{$datum} = $newvalue if @_ > 1; |
| 1409 | \& return $ClassData{$datum}; |
| 1410 | \& } |
| 1411 | \& } |
| 1412 | .Ve |
| 1413 | .PP |
| 1414 | .Vb 1 |
| 1415 | \& } |
| 1416 | .Ve |
| 1417 | .PP |
| 1418 | The closure above can be modified to take inheritance into account using |
| 1419 | the &UNIVERSAL::can method and \s-1SUPER\s0 as shown previously. |
| 1420 | .Sh "Translucency Revisited" |
| 1421 | .IX Subsection "Translucency Revisited" |
| 1422 | The Vermin class demonstrates translucency using a package variable, |
| 1423 | eponymously named \f(CW%Vermin\fR, as its meta\-object. If you prefer to |
| 1424 | use absolutely no package variables beyond those necessary to appease |
| 1425 | inheritance or possibly the Exporter, this strategy is closed to you. |
| 1426 | That's too bad, because translucent attributes are an appealing |
| 1427 | technique, so it would be valuable to devise an implementation using |
| 1428 | only lexicals. |
| 1429 | .PP |
| 1430 | There's a second reason why you might wish to avoid the eponymous |
| 1431 | package hash. If you use class names with double-colons in them, you |
| 1432 | would end up poking around somewhere you might not have meant to poke. |
| 1433 | .PP |
| 1434 | .Vb 4 |
| 1435 | \& package Vermin; |
| 1436 | \& $class = "Vermin"; |
| 1437 | \& $class->{PopCount}++; |
| 1438 | \& # accesses $Vermin::Vermin{PopCount} |
| 1439 | .Ve |
| 1440 | .PP |
| 1441 | .Vb 4 |
| 1442 | \& package Vermin::Noxious; |
| 1443 | \& $class = "Vermin::Noxious"; |
| 1444 | \& $class->{PopCount}++; |
| 1445 | \& # accesses $Vermin::Noxious{PopCount} |
| 1446 | .Ve |
| 1447 | .PP |
| 1448 | In the first case, because the class name had no double\-colons, we got |
| 1449 | the hash in the current package. But in the second case, instead of |
| 1450 | getting some hash in the current package, we got the hash \f(CW%Noxious\fR in |
| 1451 | the Vermin package. (The noxious vermin just invaded another package and |
| 1452 | sprayed their data around it. :\-) Perl doesn't support relative packages |
| 1453 | in its naming conventions, so any double-colons trigger a fully-qualified |
| 1454 | lookup instead of just looking in the current package. |
| 1455 | .PP |
| 1456 | In practice, it is unlikely that the Vermin class had an existing |
| 1457 | package variable named \f(CW%Noxious\fR that you just blew away. If you're |
| 1458 | still mistrustful, you could always stake out your own territory |
| 1459 | where you know the rules, such as using Eponymous::Vermin::Noxious or |
| 1460 | Hieronymus::Vermin::Boschious or Leave_Me_Alone::Vermin::Noxious as class |
| 1461 | names instead. Sure, it's in theory possible that someone else has |
| 1462 | a class named Eponymous::Vermin with its own \f(CW%Noxious\fR hash, but this |
| 1463 | kind of thing is always true. There's no arbiter of package names. |
| 1464 | It's always the case that globals like \f(CW@Cwd::ISA\fR would collide if more |
| 1465 | than one class uses the same Cwd package. |
| 1466 | .PP |
| 1467 | If this still leaves you with an uncomfortable twinge of paranoia, |
| 1468 | we have another solution for you. There's nothing that says that you |
| 1469 | have to have a package variable to hold a class meta\-object, either for |
| 1470 | monadic classes or for translucent attributes. Just code up the methods |
| 1471 | so that they access a lexical instead. |
| 1472 | .PP |
| 1473 | Here's another implementation of the Vermin class with semantics identical |
| 1474 | to those given previously, but this time using no package variables. |
| 1475 | .PP |
| 1476 | .Vb 1 |
| 1477 | \& package Vermin; |
| 1478 | .Ve |
| 1479 | .PP |
| 1480 | .Vb 8 |
| 1481 | \& # Here's the class meta-object, eponymously named. |
| 1482 | \& # It holds all class data, and also all instance data |
| 1483 | \& # so the latter can be used for both initialization |
| 1484 | \& # and translucency. it's a template. |
| 1485 | \& my %ClassData = ( |
| 1486 | \& PopCount => 0, # capital for class attributes |
| 1487 | \& color => "beige", # small for instance attributes |
| 1488 | \& ); |
| 1489 | .Ve |
| 1490 | .PP |
| 1491 | .Vb 13 |
| 1492 | \& # constructor method |
| 1493 | \& # invoked as class method or object method |
| 1494 | \& sub spawn { |
| 1495 | \& my $obclass = shift; |
| 1496 | \& my $class = ref($obclass) || $obclass; |
| 1497 | \& my $self = {}; |
| 1498 | \& bless($self, $class); |
| 1499 | \& $ClassData{PopCount}++; |
| 1500 | \& # init fields from invoking object, or omit if |
| 1501 | \& # invoking object is the class to provide translucency |
| 1502 | \& %$self = %$obclass if ref $obclass; |
| 1503 | \& return $self; |
| 1504 | \& } |
| 1505 | .Ve |
| 1506 | .PP |
| 1507 | .Vb 4 |
| 1508 | \& # translucent accessor for "color" attribute |
| 1509 | \& # invoked as class method or object method |
| 1510 | \& sub color { |
| 1511 | \& my $self = shift; |
| 1512 | .Ve |
| 1513 | .PP |
| 1514 | .Vb 5 |
| 1515 | \& # handle class invocation |
| 1516 | \& unless (ref $self) { |
| 1517 | \& $ClassData{color} = shift if @_; |
| 1518 | \& return $ClassData{color} |
| 1519 | \& } |
| 1520 | .Ve |
| 1521 | .PP |
| 1522 | .Vb 8 |
| 1523 | \& # handle object invocation |
| 1524 | \& $self->{color} = shift if @_; |
| 1525 | \& if (defined $self->{color}) { # not exists! |
| 1526 | \& return $self->{color}; |
| 1527 | \& } else { |
| 1528 | \& return $ClassData{color}; |
| 1529 | \& } |
| 1530 | \& } |
| 1531 | .Ve |
| 1532 | .PP |
| 1533 | .Vb 5 |
| 1534 | \& # class attribute accessor for "PopCount" attribute |
| 1535 | \& # invoked as class method or object method |
| 1536 | \& sub population { |
| 1537 | \& return $ClassData{PopCount}; |
| 1538 | \& } |
| 1539 | .Ve |
| 1540 | .PP |
| 1541 | .Vb 4 |
| 1542 | \& # instance destructor; invoked only as object method |
| 1543 | \& sub DESTROY { |
| 1544 | \& $ClassData{PopCount}--; |
| 1545 | \& } |
| 1546 | .Ve |
| 1547 | .PP |
| 1548 | .Vb 7 |
| 1549 | \& # detect whether an object attribute is translucent |
| 1550 | \& # (typically?) invoked only as object method |
| 1551 | \& sub is_translucent { |
| 1552 | \& my($self, $attr) = @_; |
| 1553 | \& $self = \e%ClassData if !ref $self; |
| 1554 | \& return !defined $self->{$attr}; |
| 1555 | \& } |
| 1556 | .Ve |
| 1557 | .PP |
| 1558 | .Vb 6 |
| 1559 | \& # test for presence of attribute in class |
| 1560 | \& # invoked as class method or object method |
| 1561 | \& sub has_attribute { |
| 1562 | \& my($self, $attr) = @_; |
| 1563 | \& return exists $ClassData{$attr}; |
| 1564 | \& } |
| 1565 | .Ve |
| 1566 | .SH "NOTES" |
| 1567 | .IX Header "NOTES" |
| 1568 | Inheritance is a powerful but subtle device, best used only after careful |
| 1569 | forethought and design. Aggregation instead of inheritance is often a |
| 1570 | better approach. |
| 1571 | .PP |
| 1572 | You can't use file-scoped lexicals in conjunction with the SelfLoader |
| 1573 | or the AutoLoader, because they alter the lexical scope in which the |
| 1574 | module's methods wind up getting compiled. |
| 1575 | .PP |
| 1576 | The usual mealy-mouthed package-mungeing doubtless applies to setting |
| 1577 | up names of object attributes. For example, \f(CW\*(C`$self\->{ObData1}\*(C'\fR |
| 1578 | should probably be \f(CW\*(C`$self\->{ _\|_PACKAGE_\|_ . "_ObData1" }\*(C'\fR, but that |
| 1579 | would just confuse the examples. |
| 1580 | .SH "SEE ALSO" |
| 1581 | .IX Header "SEE ALSO" |
| 1582 | perltoot, perlobj, perlmod, and perlbot. |
| 1583 | .PP |
| 1584 | The Tie::SecureHash and Class::Data::Inheritable modules from \s-1CPAN\s0 are |
| 1585 | worth checking out. |
| 1586 | .SH "AUTHOR AND COPYRIGHT" |
| 1587 | .IX Header "AUTHOR AND COPYRIGHT" |
| 1588 | Copyright (c) 1999 Tom Christiansen. |
| 1589 | All rights reserved. |
| 1590 | .PP |
| 1591 | This documentation is free; you can redistribute it and/or modify it |
| 1592 | under the same terms as Perl itself. |
| 1593 | .PP |
| 1594 | Irrespective of its distribution, all code examples in this file |
| 1595 | are hereby placed into the public domain. You are permitted and |
| 1596 | encouraged to use this code in your own programs for fun |
| 1597 | or for profit as you see fit. A simple comment in the code giving |
| 1598 | credit would be courteous but is not required. |
| 1599 | .SH "ACKNOWLEDGEMENTS" |
| 1600 | .IX Header "ACKNOWLEDGEMENTS" |
| 1601 | Russ Allbery, Jon Orwant, Randy Ray, Larry Rosler, Nat Torkington, |
| 1602 | and Stephen Warren all contributed suggestions and corrections to this |
| 1603 | piece. Thanks especially to Damian Conway for his ideas and feedback, |
| 1604 | and without whose indirect prodding I might never have taken the time |
| 1605 | to show others how much Perl has to offer in the way of objects once |
| 1606 | you start thinking outside the tiny little box that today's \*(L"popular\*(R" |
| 1607 | object-oriented languages enforce. |
| 1608 | .SH "HISTORY" |
| 1609 | .IX Header "HISTORY" |
| 1610 | Last edit: Sun Feb 4 20:50:28 \s-1EST\s0 2001 |