| 1 | =head1 NAME |
| 2 | |
| 3 | perlbot - Bag'o Object Tricks (the BOT) |
| 4 | |
| 5 | =head1 DESCRIPTION |
| 6 | |
| 7 | The following collection of tricks and hints is intended to whet curious |
| 8 | appetites about such things as the use of instance variables and the |
| 9 | mechanics of object and class relationships. The reader is encouraged to |
| 10 | consult relevant textbooks for discussion of Object Oriented definitions and |
| 11 | methodology. This is not intended as a tutorial for object-oriented |
| 12 | programming or as a comprehensive guide to Perl's object oriented features, |
| 13 | nor should it be construed as a style guide. If you're looking for tutorials, |
| 14 | be sure to read L<perlboot>, L<perltoot>, and L<perltooc>. |
| 15 | |
| 16 | The Perl motto still holds: There's more than one way to do it. |
| 17 | |
| 18 | =head1 OO SCALING TIPS |
| 19 | |
| 20 | =over 5 |
| 21 | |
| 22 | =item 1 |
| 23 | |
| 24 | Do not attempt to verify the type of $self. That'll break if the class is |
| 25 | inherited, when the type of $self is valid but its package isn't what you |
| 26 | expect. See rule 5. |
| 27 | |
| 28 | =item 2 |
| 29 | |
| 30 | If an object-oriented (OO) or indirect-object (IO) syntax was used, then the |
| 31 | object is probably the correct type and there's no need to become paranoid |
| 32 | about it. Perl isn't a paranoid language anyway. If people subvert the OO |
| 33 | or IO syntax then they probably know what they're doing and you should let |
| 34 | them do it. See rule 1. |
| 35 | |
| 36 | =item 3 |
| 37 | |
| 38 | Use the two-argument form of bless(). Let a subclass use your constructor. |
| 39 | See L<INHERITING A CONSTRUCTOR>. |
| 40 | |
| 41 | =item 4 |
| 42 | |
| 43 | The subclass is allowed to know things about its immediate superclass, the |
| 44 | superclass is allowed to know nothing about a subclass. |
| 45 | |
| 46 | =item 5 |
| 47 | |
| 48 | Don't be trigger happy with inheritance. A "using", "containing", or |
| 49 | "delegation" relationship (some sort of aggregation, at least) is often more |
| 50 | appropriate. See L<OBJECT RELATIONSHIPS>, L<USING RELATIONSHIP WITH SDBM>, |
| 51 | and L<"DELEGATION">. |
| 52 | |
| 53 | =item 6 |
| 54 | |
| 55 | The object is the namespace. Make package globals accessible via the |
| 56 | object. This will remove the guess work about the symbol's home package. |
| 57 | See L<CLASS CONTEXT AND THE OBJECT>. |
| 58 | |
| 59 | =item 7 |
| 60 | |
| 61 | IO syntax is certainly less noisy, but it is also prone to ambiguities that |
| 62 | can cause difficult-to-find bugs. Allow people to use the sure-thing OO |
| 63 | syntax, even if you don't like it. |
| 64 | |
| 65 | =item 8 |
| 66 | |
| 67 | Do not use function-call syntax on a method. You're going to be bitten |
| 68 | someday. Someone might move that method into a superclass and your code |
| 69 | will be broken. On top of that you're feeding the paranoia in rule 2. |
| 70 | |
| 71 | =item 9 |
| 72 | |
| 73 | Don't assume you know the home package of a method. You're making it |
| 74 | difficult for someone to override that method. See L<THINKING OF CODE REUSE>. |
| 75 | |
| 76 | =back |
| 77 | |
| 78 | =head1 INSTANCE VARIABLES |
| 79 | |
| 80 | An anonymous array or anonymous hash can be used to hold instance |
| 81 | variables. Named parameters are also demonstrated. |
| 82 | |
| 83 | package Foo; |
| 84 | |
| 85 | sub new { |
| 86 | my $type = shift; |
| 87 | my %params = @_; |
| 88 | my $self = {}; |
| 89 | $self->{'High'} = $params{'High'}; |
| 90 | $self->{'Low'} = $params{'Low'}; |
| 91 | bless $self, $type; |
| 92 | } |
| 93 | |
| 94 | |
| 95 | package Bar; |
| 96 | |
| 97 | sub new { |
| 98 | my $type = shift; |
| 99 | my %params = @_; |
| 100 | my $self = []; |
| 101 | $self->[0] = $params{'Left'}; |
| 102 | $self->[1] = $params{'Right'}; |
| 103 | bless $self, $type; |
| 104 | } |
| 105 | |
| 106 | package main; |
| 107 | |
| 108 | $a = Foo->new( 'High' => 42, 'Low' => 11 ); |
| 109 | print "High=$a->{'High'}\n"; |
| 110 | print "Low=$a->{'Low'}\n"; |
| 111 | |
| 112 | $b = Bar->new( 'Left' => 78, 'Right' => 40 ); |
| 113 | print "Left=$b->[0]\n"; |
| 114 | print "Right=$b->[1]\n"; |
| 115 | |
| 116 | =head1 SCALAR INSTANCE VARIABLES |
| 117 | |
| 118 | An anonymous scalar can be used when only one instance variable is needed. |
| 119 | |
| 120 | package Foo; |
| 121 | |
| 122 | sub new { |
| 123 | my $type = shift; |
| 124 | my $self; |
| 125 | $self = shift; |
| 126 | bless \$self, $type; |
| 127 | } |
| 128 | |
| 129 | package main; |
| 130 | |
| 131 | $a = Foo->new( 42 ); |
| 132 | print "a=$$a\n"; |
| 133 | |
| 134 | |
| 135 | =head1 INSTANCE VARIABLE INHERITANCE |
| 136 | |
| 137 | This example demonstrates how one might inherit instance variables from a |
| 138 | superclass for inclusion in the new class. This requires calling the |
| 139 | superclass's constructor and adding one's own instance variables to the new |
| 140 | object. |
| 141 | |
| 142 | package Bar; |
| 143 | |
| 144 | sub new { |
| 145 | my $type = shift; |
| 146 | my $self = {}; |
| 147 | $self->{'buz'} = 42; |
| 148 | bless $self, $type; |
| 149 | } |
| 150 | |
| 151 | package Foo; |
| 152 | @ISA = qw( Bar ); |
| 153 | |
| 154 | sub new { |
| 155 | my $type = shift; |
| 156 | my $self = Bar->new; |
| 157 | $self->{'biz'} = 11; |
| 158 | bless $self, $type; |
| 159 | } |
| 160 | |
| 161 | package main; |
| 162 | |
| 163 | $a = Foo->new; |
| 164 | print "buz = ", $a->{'buz'}, "\n"; |
| 165 | print "biz = ", $a->{'biz'}, "\n"; |
| 166 | |
| 167 | |
| 168 | |
| 169 | =head1 OBJECT RELATIONSHIPS |
| 170 | |
| 171 | The following demonstrates how one might implement "containing" and "using" |
| 172 | relationships between objects. |
| 173 | |
| 174 | package Bar; |
| 175 | |
| 176 | sub new { |
| 177 | my $type = shift; |
| 178 | my $self = {}; |
| 179 | $self->{'buz'} = 42; |
| 180 | bless $self, $type; |
| 181 | } |
| 182 | |
| 183 | package Foo; |
| 184 | |
| 185 | sub new { |
| 186 | my $type = shift; |
| 187 | my $self = {}; |
| 188 | $self->{'Bar'} = Bar->new; |
| 189 | $self->{'biz'} = 11; |
| 190 | bless $self, $type; |
| 191 | } |
| 192 | |
| 193 | package main; |
| 194 | |
| 195 | $a = Foo->new; |
| 196 | print "buz = ", $a->{'Bar'}->{'buz'}, "\n"; |
| 197 | print "biz = ", $a->{'biz'}, "\n"; |
| 198 | |
| 199 | |
| 200 | |
| 201 | =head1 OVERRIDING SUPERCLASS METHODS |
| 202 | |
| 203 | The following example demonstrates how to override a superclass method and |
| 204 | then call the overridden method. The B<SUPER> pseudo-class allows the |
| 205 | programmer to call an overridden superclass method without actually knowing |
| 206 | where that method is defined. |
| 207 | |
| 208 | package Buz; |
| 209 | sub goo { print "here's the goo\n" } |
| 210 | |
| 211 | package Bar; @ISA = qw( Buz ); |
| 212 | sub google { print "google here\n" } |
| 213 | |
| 214 | package Baz; |
| 215 | sub mumble { print "mumbling\n" } |
| 216 | |
| 217 | package Foo; |
| 218 | @ISA = qw( Bar Baz ); |
| 219 | |
| 220 | sub new { |
| 221 | my $type = shift; |
| 222 | bless [], $type; |
| 223 | } |
| 224 | sub grr { print "grumble\n" } |
| 225 | sub goo { |
| 226 | my $self = shift; |
| 227 | $self->SUPER::goo(); |
| 228 | } |
| 229 | sub mumble { |
| 230 | my $self = shift; |
| 231 | $self->SUPER::mumble(); |
| 232 | } |
| 233 | sub google { |
| 234 | my $self = shift; |
| 235 | $self->SUPER::google(); |
| 236 | } |
| 237 | |
| 238 | package main; |
| 239 | |
| 240 | $foo = Foo->new; |
| 241 | $foo->mumble; |
| 242 | $foo->grr; |
| 243 | $foo->goo; |
| 244 | $foo->google; |
| 245 | |
| 246 | Note that C<SUPER> refers to the superclasses of the current package |
| 247 | (C<Foo>), not to the superclasses of C<$self>. |
| 248 | |
| 249 | |
| 250 | =head1 USING RELATIONSHIP WITH SDBM |
| 251 | |
| 252 | This example demonstrates an interface for the SDBM class. This creates a |
| 253 | "using" relationship between the SDBM class and the new class Mydbm. |
| 254 | |
| 255 | package Mydbm; |
| 256 | |
| 257 | require SDBM_File; |
| 258 | require Tie::Hash; |
| 259 | @ISA = qw( Tie::Hash ); |
| 260 | |
| 261 | sub TIEHASH { |
| 262 | my $type = shift; |
| 263 | my $ref = SDBM_File->new(@_); |
| 264 | bless {'dbm' => $ref}, $type; |
| 265 | } |
| 266 | sub FETCH { |
| 267 | my $self = shift; |
| 268 | my $ref = $self->{'dbm'}; |
| 269 | $ref->FETCH(@_); |
| 270 | } |
| 271 | sub STORE { |
| 272 | my $self = shift; |
| 273 | if (defined $_[0]){ |
| 274 | my $ref = $self->{'dbm'}; |
| 275 | $ref->STORE(@_); |
| 276 | } else { |
| 277 | die "Cannot STORE an undefined key in Mydbm\n"; |
| 278 | } |
| 279 | } |
| 280 | |
| 281 | package main; |
| 282 | use Fcntl qw( O_RDWR O_CREAT ); |
| 283 | |
| 284 | tie %foo, "Mydbm", "Sdbm", O_RDWR|O_CREAT, 0640; |
| 285 | $foo{'bar'} = 123; |
| 286 | print "foo-bar = $foo{'bar'}\n"; |
| 287 | |
| 288 | tie %bar, "Mydbm", "Sdbm2", O_RDWR|O_CREAT, 0640; |
| 289 | $bar{'Cathy'} = 456; |
| 290 | print "bar-Cathy = $bar{'Cathy'}\n"; |
| 291 | |
| 292 | =head1 THINKING OF CODE REUSE |
| 293 | |
| 294 | One strength of Object-Oriented languages is the ease with which old code |
| 295 | can use new code. The following examples will demonstrate first how one can |
| 296 | hinder code reuse and then how one can promote code reuse. |
| 297 | |
| 298 | This first example illustrates a class which uses a fully-qualified method |
| 299 | call to access the "private" method BAZ(). The second example will show |
| 300 | that it is impossible to override the BAZ() method. |
| 301 | |
| 302 | package FOO; |
| 303 | |
| 304 | sub new { |
| 305 | my $type = shift; |
| 306 | bless {}, $type; |
| 307 | } |
| 308 | sub bar { |
| 309 | my $self = shift; |
| 310 | $self->FOO::private::BAZ; |
| 311 | } |
| 312 | |
| 313 | package FOO::private; |
| 314 | |
| 315 | sub BAZ { |
| 316 | print "in BAZ\n"; |
| 317 | } |
| 318 | |
| 319 | package main; |
| 320 | |
| 321 | $a = FOO->new; |
| 322 | $a->bar; |
| 323 | |
| 324 | Now we try to override the BAZ() method. We would like FOO::bar() to call |
| 325 | GOOP::BAZ(), but this cannot happen because FOO::bar() explicitly calls |
| 326 | FOO::private::BAZ(). |
| 327 | |
| 328 | package FOO; |
| 329 | |
| 330 | sub new { |
| 331 | my $type = shift; |
| 332 | bless {}, $type; |
| 333 | } |
| 334 | sub bar { |
| 335 | my $self = shift; |
| 336 | $self->FOO::private::BAZ; |
| 337 | } |
| 338 | |
| 339 | package FOO::private; |
| 340 | |
| 341 | sub BAZ { |
| 342 | print "in BAZ\n"; |
| 343 | } |
| 344 | |
| 345 | package GOOP; |
| 346 | @ISA = qw( FOO ); |
| 347 | sub new { |
| 348 | my $type = shift; |
| 349 | bless {}, $type; |
| 350 | } |
| 351 | |
| 352 | sub BAZ { |
| 353 | print "in GOOP::BAZ\n"; |
| 354 | } |
| 355 | |
| 356 | package main; |
| 357 | |
| 358 | $a = GOOP->new; |
| 359 | $a->bar; |
| 360 | |
| 361 | To create reusable code we must modify class FOO, flattening class |
| 362 | FOO::private. The next example shows a reusable class FOO which allows the |
| 363 | method GOOP::BAZ() to be used in place of FOO::BAZ(). |
| 364 | |
| 365 | package FOO; |
| 366 | |
| 367 | sub new { |
| 368 | my $type = shift; |
| 369 | bless {}, $type; |
| 370 | } |
| 371 | sub bar { |
| 372 | my $self = shift; |
| 373 | $self->BAZ; |
| 374 | } |
| 375 | |
| 376 | sub BAZ { |
| 377 | print "in BAZ\n"; |
| 378 | } |
| 379 | |
| 380 | package GOOP; |
| 381 | @ISA = qw( FOO ); |
| 382 | |
| 383 | sub new { |
| 384 | my $type = shift; |
| 385 | bless {}, $type; |
| 386 | } |
| 387 | sub BAZ { |
| 388 | print "in GOOP::BAZ\n"; |
| 389 | } |
| 390 | |
| 391 | package main; |
| 392 | |
| 393 | $a = GOOP->new; |
| 394 | $a->bar; |
| 395 | |
| 396 | =head1 CLASS CONTEXT AND THE OBJECT |
| 397 | |
| 398 | Use the object to solve package and class context problems. Everything a |
| 399 | method needs should be available via the object or should be passed as a |
| 400 | parameter to the method. |
| 401 | |
| 402 | A class will sometimes have static or global data to be used by the |
| 403 | methods. A subclass may want to override that data and replace it with new |
| 404 | data. When this happens the superclass may not know how to find the new |
| 405 | copy of the data. |
| 406 | |
| 407 | This problem can be solved by using the object to define the context of the |
| 408 | method. Let the method look in the object for a reference to the data. The |
| 409 | alternative is to force the method to go hunting for the data ("Is it in my |
| 410 | class, or in a subclass? Which subclass?"), and this can be inconvenient |
| 411 | and will lead to hackery. It is better just to let the object tell the |
| 412 | method where that data is located. |
| 413 | |
| 414 | package Bar; |
| 415 | |
| 416 | %fizzle = ( 'Password' => 'XYZZY' ); |
| 417 | |
| 418 | sub new { |
| 419 | my $type = shift; |
| 420 | my $self = {}; |
| 421 | $self->{'fizzle'} = \%fizzle; |
| 422 | bless $self, $type; |
| 423 | } |
| 424 | |
| 425 | sub enter { |
| 426 | my $self = shift; |
| 427 | |
| 428 | # Don't try to guess if we should use %Bar::fizzle |
| 429 | # or %Foo::fizzle. The object already knows which |
| 430 | # we should use, so just ask it. |
| 431 | # |
| 432 | my $fizzle = $self->{'fizzle'}; |
| 433 | |
| 434 | print "The word is ", $fizzle->{'Password'}, "\n"; |
| 435 | } |
| 436 | |
| 437 | package Foo; |
| 438 | @ISA = qw( Bar ); |
| 439 | |
| 440 | %fizzle = ( 'Password' => 'Rumple' ); |
| 441 | |
| 442 | sub new { |
| 443 | my $type = shift; |
| 444 | my $self = Bar->new; |
| 445 | $self->{'fizzle'} = \%fizzle; |
| 446 | bless $self, $type; |
| 447 | } |
| 448 | |
| 449 | package main; |
| 450 | |
| 451 | $a = Bar->new; |
| 452 | $b = Foo->new; |
| 453 | $a->enter; |
| 454 | $b->enter; |
| 455 | |
| 456 | =head1 INHERITING A CONSTRUCTOR |
| 457 | |
| 458 | An inheritable constructor should use the second form of bless() which allows |
| 459 | blessing directly into a specified class. Notice in this example that the |
| 460 | object will be a BAR not a FOO, even though the constructor is in class FOO. |
| 461 | |
| 462 | package FOO; |
| 463 | |
| 464 | sub new { |
| 465 | my $type = shift; |
| 466 | my $self = {}; |
| 467 | bless $self, $type; |
| 468 | } |
| 469 | |
| 470 | sub baz { |
| 471 | print "in FOO::baz()\n"; |
| 472 | } |
| 473 | |
| 474 | package BAR; |
| 475 | @ISA = qw(FOO); |
| 476 | |
| 477 | sub baz { |
| 478 | print "in BAR::baz()\n"; |
| 479 | } |
| 480 | |
| 481 | package main; |
| 482 | |
| 483 | $a = BAR->new; |
| 484 | $a->baz; |
| 485 | |
| 486 | =head1 DELEGATION |
| 487 | |
| 488 | Some classes, such as SDBM_File, cannot be effectively subclassed because |
| 489 | they create foreign objects. Such a class can be extended with some sort of |
| 490 | aggregation technique such as the "using" relationship mentioned earlier or |
| 491 | by delegation. |
| 492 | |
| 493 | The following example demonstrates delegation using an AUTOLOAD() function to |
| 494 | perform message-forwarding. This will allow the Mydbm object to behave |
| 495 | exactly like an SDBM_File object. The Mydbm class could now extend the |
| 496 | behavior by adding custom FETCH() and STORE() methods, if this is desired. |
| 497 | |
| 498 | package Mydbm; |
| 499 | |
| 500 | require SDBM_File; |
| 501 | require Tie::Hash; |
| 502 | @ISA = qw(Tie::Hash); |
| 503 | |
| 504 | sub TIEHASH { |
| 505 | my $type = shift; |
| 506 | my $ref = SDBM_File->new(@_); |
| 507 | bless {'delegate' => $ref}; |
| 508 | } |
| 509 | |
| 510 | sub AUTOLOAD { |
| 511 | my $self = shift; |
| 512 | |
| 513 | # The Perl interpreter places the name of the |
| 514 | # message in a variable called $AUTOLOAD. |
| 515 | |
| 516 | # DESTROY messages should never be propagated. |
| 517 | return if $AUTOLOAD =~ /::DESTROY$/; |
| 518 | |
| 519 | # Remove the package name. |
| 520 | $AUTOLOAD =~ s/^Mydbm:://; |
| 521 | |
| 522 | # Pass the message to the delegate. |
| 523 | $self->{'delegate'}->$AUTOLOAD(@_); |
| 524 | } |
| 525 | |
| 526 | package main; |
| 527 | use Fcntl qw( O_RDWR O_CREAT ); |
| 528 | |
| 529 | tie %foo, "Mydbm", "adbm", O_RDWR|O_CREAT, 0640; |
| 530 | $foo{'bar'} = 123; |
| 531 | print "foo-bar = $foo{'bar'}\n"; |
| 532 | |
| 533 | =head1 SEE ALSO |
| 534 | |
| 535 | L<perlboot>, L<perltoot>, L<perltooc>. |