| 1 | @macro goops |
| 2 | GOOPS |
| 3 | @end macro |
| 4 | |
| 5 | @macro guile |
| 6 | Guile |
| 7 | @end macro |
| 8 | |
| 9 | @node GOOPS |
| 10 | @chapter GOOPS |
| 11 | |
| 12 | @menu |
| 13 | * Introduction:: |
| 14 | * Getting Started:: |
| 15 | * Reference Manual:: |
| 16 | * MOP Specification:: |
| 17 | * Tutorial:: |
| 18 | @end menu |
| 19 | |
| 20 | @node Introduction |
| 21 | @section Introduction |
| 22 | |
| 23 | @goops{} is the object oriented extension to @guile{}. Its |
| 24 | implementation is derived from @w{STk-3.99.3} by Erick Gallesio and |
| 25 | version 1.3 of Gregor Kiczales @cite{Tiny-Clos}. It is very close in |
| 26 | spirit to CLOS, the Common Lisp Object System (@cite{CLtL2}) but is |
| 27 | adapted for the Scheme language. While GOOPS is not compatible with any |
| 28 | of these systems, GOOPS contains a compatibility module which allows for |
| 29 | execution of STKlos programs. |
| 30 | |
| 31 | Briefly stated, the @goops{} extension gives the user a full object |
| 32 | oriented system with multiple inheritance and generic functions with |
| 33 | multi-method dispatch. Furthermore, the implementation relies on a true |
| 34 | meta object protocol, in the spirit of the one defined for CLOS |
| 35 | (@cite{Gregor Kiczales: A Metaobject Protocol}). |
| 36 | |
| 37 | @node Getting Started |
| 38 | @section Getting Started |
| 39 | |
| 40 | @menu |
| 41 | * Running GOOPS:: |
| 42 | |
| 43 | Examples of some basic GOOPS functionality. |
| 44 | |
| 45 | * Methods:: |
| 46 | * User-defined types:: |
| 47 | * Asking for the type of an object:: |
| 48 | |
| 49 | See further in the GOOPS tutorial available in this distribution in |
| 50 | info (goops.info) and texinfo format. |
| 51 | @end menu |
| 52 | |
| 53 | @node Running GOOPS |
| 54 | @subsection Running GOOPS |
| 55 | |
| 56 | @enumerate |
| 57 | @item |
| 58 | Type |
| 59 | |
| 60 | @smalllisp |
| 61 | guile-oops |
| 62 | @end smalllisp |
| 63 | |
| 64 | You should now be at the Guile prompt ("guile> "). |
| 65 | |
| 66 | @item |
| 67 | Type |
| 68 | |
| 69 | @smalllisp |
| 70 | (use-modules (oop goops)) |
| 71 | @end smalllisp |
| 72 | |
| 73 | to load GOOPS. (If your system supports dynamic loading, you |
| 74 | should be able to do this not only from `guile-oops' but from an |
| 75 | arbitrary Guile interpreter.) |
| 76 | @end enumerate |
| 77 | |
| 78 | We're now ready to try some basic GOOPS functionality. |
| 79 | |
| 80 | @node Methods |
| 81 | @subsection Methods |
| 82 | |
| 83 | @smalllisp |
| 84 | @group |
| 85 | (define-method (+ (x <string>) (y <string>)) |
| 86 | (string-append x y)) |
| 87 | |
| 88 | (+ 1 2) --> 3 |
| 89 | (+ "abc" "de") --> "abcde" |
| 90 | @end group |
| 91 | @end smalllisp |
| 92 | |
| 93 | @node User-defined types |
| 94 | @subsection User-defined types |
| 95 | |
| 96 | @smalllisp |
| 97 | (define-class <2D-vector> () |
| 98 | (x #:init-value 0 #:accessor x-component #:init-keyword #:x) |
| 99 | (y #:init-value 0 #:accessor y-component #:init-keyword #:y)) |
| 100 | |
| 101 | @group |
| 102 | (use-modules (ice-9 format)) |
| 103 | |
| 104 | (define-method (write (obj <2D-vector>) port) |
| 105 | (display (format #f "<~S, ~S>" (x-component obj) (y-component obj)) |
| 106 | port)) |
| 107 | |
| 108 | (define v (make <2D-vector> #:x 3 #:y 4)) |
| 109 | |
| 110 | v --> <3, 4> |
| 111 | @end group |
| 112 | |
| 113 | @group |
| 114 | (define-method (+ (x <2D-vector>) (y <2D-vector>)) |
| 115 | (make <2D-vector> |
| 116 | #:x (+ (x-component x) (x-component y)) |
| 117 | #:y (+ (y-component x) (y-component y)))) |
| 118 | |
| 119 | (+ v v) --> <6, 8> |
| 120 | @end group |
| 121 | @end smalllisp |
| 122 | |
| 123 | @node Asking for the type of an object |
| 124 | @subsection Types |
| 125 | |
| 126 | @example |
| 127 | (class-of v) --> #<<class> <2D-vector> 40241ac0> |
| 128 | <2D-vector> --> #<<class> <2D-vector> 40241ac0> |
| 129 | (class-of 1) --> #<<class> <integer> 401b2a98> |
| 130 | <integer> --> #<<class> <integer> 401b2a98> |
| 131 | |
| 132 | (is-a? v <2D-vector>) --> #t |
| 133 | @end example |
| 134 | |
| 135 | @node Reference Manual |
| 136 | @section Reference Manual |
| 137 | |
| 138 | This chapter is the GOOPS reference manual. It aims to describe all the |
| 139 | syntax, procedures, options and associated concepts that a typical |
| 140 | application author would need to understand in order to use GOOPS |
| 141 | effectively in their application. It also describes what is meant by |
| 142 | the GOOPS ``metaobject protocol'' (aka ``MOP''), and indicates how |
| 143 | authors can use the metaobject protocol to customize the behaviour of |
| 144 | GOOPS itself. |
| 145 | |
| 146 | For a detailed specification of the GOOPS metaobject protocol, see |
| 147 | @ref{MOP Specification}. |
| 148 | |
| 149 | @menu |
| 150 | * Introductory Remarks:: |
| 151 | * Defining New Classes:: |
| 152 | * Creating Instances:: |
| 153 | * Accessing Slots:: |
| 154 | * Creating Generic Functions:: |
| 155 | * Adding Methods to Generic Functions:: |
| 156 | * Invoking Generic Functions:: |
| 157 | * Redefining a Class:: |
| 158 | * Changing the Class of an Instance:: |
| 159 | * Introspection:: |
| 160 | * Miscellaneous Functions:: |
| 161 | @end menu |
| 162 | |
| 163 | @node Introductory Remarks |
| 164 | @subsection Introductory Remarks |
| 165 | |
| 166 | GOOPS is an object-oriented programming system based on a ``metaobject |
| 167 | protocol'' derived from the ones used in CLOS (the Common Lisp Object |
| 168 | System), tiny-clos (a small Scheme implementation of a subset of CLOS |
| 169 | functionality) and STKlos. |
| 170 | |
| 171 | GOOPS can be used by application authors at a basic level without any |
| 172 | need to understand what the metaobject protocol (aka ``MOP'') is and how |
| 173 | it works. On the other hand, the MOP underlies even the customizations |
| 174 | that application authors are likely to make use of very quickly --- such |
| 175 | as defining an @code{initialize} method to customize the initialization |
| 176 | of instances of an application-defined class --- and an understanding of |
| 177 | the MOP makes it much easier to explain such customizations in a precise |
| 178 | way. And in the long run, understanding the MOP is the key both to |
| 179 | understanding GOOPS at a deeper level and to taking full advantage of |
| 180 | GOOPS' power, by customizing the behaviour of GOOPS itself. |
| 181 | |
| 182 | Each of the following sections of the reference manual is arranged |
| 183 | such that the most basic usage is introduced first, and then subsequent |
| 184 | subsubsections discuss the related internal functions and metaobject |
| 185 | protocols, finishing with a description of how to customize that area of |
| 186 | functionality. |
| 187 | |
| 188 | These introductory remarks continue with a few words about metaobjects |
| 189 | and the MOP. Readers who do not want to be bothered yet with the MOP |
| 190 | and customization could safely skip this subsubsection on a first reading, |
| 191 | and should correspondingly skip subsequent subsubsections that are |
| 192 | concerned with internals and customization. |
| 193 | |
| 194 | In general, this reference manual assumes familiarity with standard |
| 195 | object oriented concepts and terminology. However, some of the terms |
| 196 | used in GOOPS are less well known, so the Terminology subsubsection |
| 197 | provides definitions for these terms. |
| 198 | |
| 199 | @menu |
| 200 | * Metaobjects and the Metaobject Protocol:: |
| 201 | * Terminology:: |
| 202 | @end menu |
| 203 | |
| 204 | @node Metaobjects and the Metaobject Protocol |
| 205 | @subsubsection Metaobjects and the Metaobject Protocol |
| 206 | |
| 207 | The conceptual building blocks of GOOPS are classes, slot definitions, |
| 208 | instances, generic functions and methods. A class is a grouping of |
| 209 | inheritance relations and slot definitions. An instance is an object |
| 210 | with slots that are allocated following the rules implied by its class's |
| 211 | superclasses and slot definitions. A generic function is a collection |
| 212 | of methods and rules for determining which of those methods to apply |
| 213 | when the generic function is invoked. A method is a procedure and a set |
| 214 | of specializers that specify the type of arguments to which the |
| 215 | procedure is applicable. |
| 216 | |
| 217 | Of these entities, GOOPS represents classes, generic functions and |
| 218 | methods as ``metaobjects''. In other words, the values in a GOOPS |
| 219 | program that describe classes, generic functions and methods, are |
| 220 | themselves instances (or ``objects'') of special GOOPS classes that |
| 221 | encapsulate the behaviour, respectively, of classes, generic functions, |
| 222 | and methods. |
| 223 | |
| 224 | (The other two entities are slot definitions and instances. Slot |
| 225 | definitions are not strictly instances, but every slot definition is |
| 226 | associated with a GOOPS class that specifies the behaviour of the slot |
| 227 | as regards accessibility and protection from garbage collection. |
| 228 | Instances are of course objects in the usual sense, and there is no |
| 229 | benefit from thinking of them as metaobjects.) |
| 230 | |
| 231 | The ``metaobject protocol'' (aka ``MOP'') is the specification of the |
| 232 | generic functions which determine the behaviour of these metaobjects and |
| 233 | the circumstances in which these generic functions are invoked. |
| 234 | |
| 235 | For a concrete example of what this means, consider how GOOPS calculates |
| 236 | the set of slots for a class that is being defined using |
| 237 | @code{define-class}. The desired set of slots is the union of the new |
| 238 | class's direct slots and the slots of all its superclasses. But |
| 239 | @code{define-class} itself does not perform this calculation. Instead, |
| 240 | there is a method of the @code{initialize} generic function that is |
| 241 | specialized for instances of type @code{<class>}, and it is this method |
| 242 | that performs the slot calculation. |
| 243 | |
| 244 | @code{initialize} is a generic function which GOOPS calls whenever a new |
| 245 | instance is created, immediately after allocating memory for a new |
| 246 | instance, in order to initialize the new instance's slots. The sequence |
| 247 | of steps is as follows. |
| 248 | |
| 249 | @itemize @bullet |
| 250 | @item |
| 251 | @code{define-class} uses @code{make} to make a new instance of the |
| 252 | @code{<class>}, passing as initialization arguments the superclasses, |
| 253 | slot definitions and class options that were specified in the |
| 254 | @code{define-class} form. |
| 255 | |
| 256 | @item |
| 257 | @code{make} allocates memory for the new instance, and then invokes the |
| 258 | @code{initialize} generic function to initialize the new instance's |
| 259 | slots. |
| 260 | |
| 261 | @item |
| 262 | The @code{initialize} generic function applies the method that is |
| 263 | specialized for instances of type @code{<class>}, and this method |
| 264 | performs the slot calculation. |
| 265 | @end itemize |
| 266 | |
| 267 | In other words, rather than being hardcoded in @code{define-class}, the |
| 268 | behaviour of class definition is encapsulated by generic function |
| 269 | methods that are specialized for the class @code{<class>}. |
| 270 | |
| 271 | It is possible to create a new class that inherits from @code{<class>}, |
| 272 | which is called a ``metaclass'', and to write a new @code{initialize} |
| 273 | method that is specialized for instances of the new metaclass. Then, if |
| 274 | the @code{define-class} form includes a @code{#:metaclass} class option |
| 275 | whose value is the new metaclass, the class that is defined by the |
| 276 | @code{define-class} form will be an instance of the new metaclass rather |
| 277 | than of the default @code{<class>}, and will be defined in accordance |
| 278 | with the new @code{initialize} method. Thus the default slot |
| 279 | calculation, as well as any other aspect of the new class's relationship |
| 280 | with its superclasses, can be modified or overridden. |
| 281 | |
| 282 | In a similar way, the behaviour of generic functions can be modified or |
| 283 | overridden by creating a new class that inherits from the standard |
| 284 | generic function class @code{<generic>}, writing appropriate methods |
| 285 | that are specialized to the new class, and creating new generic |
| 286 | functions that are instances of the new class. |
| 287 | |
| 288 | The same is true for method metaobjects. And the same basic mechanism |
| 289 | allows the application class author to write an @code{initialize} method |
| 290 | that is specialized to their application class, to initialize instances |
| 291 | of that class. |
| 292 | |
| 293 | Such is the power of the MOP. Note that @code{initialize} is just one |
| 294 | of a large number of generic functions that can be customized to modify |
| 295 | the behaviour of application objects and classes and of GOOPS itself. |
| 296 | Each subsequent section of the reference manual covers a particular area |
| 297 | of GOOPS functionality, and describes the generic functions that are |
| 298 | relevant for customization of that area. |
| 299 | |
| 300 | We conclude this subsubsection by emphasizing a point that may seem |
| 301 | obvious, but contrasts with the corresponding situation in some other |
| 302 | MOP implementations, such as CLOS. The point is simply that an |
| 303 | identifier which represents a GOOPS class or generic function is a |
| 304 | variable with a first-class value, the value being an instance of class |
| 305 | @code{<class>} or @code{<generic>}. (In CLOS, on the other hand, a |
| 306 | class identifier is a symbol that indexes the corresponding class |
| 307 | metaobject in a separate namespace for classes.) This is, of course, |
| 308 | simply an extension of the tendency in Scheme to avoid the unnecessary |
| 309 | use of, on the one hand, syntactic forms that require unevaluated |
| 310 | arguments and, on the other, separate identifier namespaces (e.g. for |
| 311 | class names), but it is worth noting that GOOPS conforms fully to this |
| 312 | Schemely principle. |
| 313 | |
| 314 | @node Terminology |
| 315 | @subsubsection Terminology |
| 316 | |
| 317 | It is assumed that the reader is already familiar with standard object |
| 318 | orientation concepts such as classes, objects/instances, |
| 319 | inheritance/subclassing, generic functions and methods, encapsulation |
| 320 | and polymorphism. |
| 321 | |
| 322 | This section explains some of the less well known concepts and |
| 323 | terminology that GOOPS uses, which are assumed by the following sections |
| 324 | of the reference manual. |
| 325 | |
| 326 | @subsubheading Metaclass |
| 327 | |
| 328 | A @dfn{metaclass} is the class of an object which represents a GOOPS |
| 329 | class. Put more succinctly, a metaclass is a class's class. |
| 330 | |
| 331 | Most GOOPS classes have the metaclass @code{<class>} and, by default, |
| 332 | any new class that is created using @code{define-class} has the |
| 333 | metaclass @code{<class>}. |
| 334 | |
| 335 | But what does this really mean? To find out, let's look in more detail |
| 336 | at what happens when a new class is created using @code{define-class}: |
| 337 | |
| 338 | @example |
| 339 | (define-class <my-class> (<object>) . slots) |
| 340 | @end example |
| 341 | |
| 342 | GOOPS actually expands the @code{define-class} form to something like |
| 343 | this |
| 344 | |
| 345 | @example |
| 346 | (define <my-class> (class (<object>) . slots)) |
| 347 | @end example |
| 348 | |
| 349 | and thence to |
| 350 | |
| 351 | @example |
| 352 | (define <my-class> |
| 353 | (make <class> #:supers (list <object>) #:slots slots)) |
| 354 | @end example |
| 355 | |
| 356 | In other words, the value of @code{<my-class>} is in fact an instance of |
| 357 | the class @code{<class>} with slot values specifying the superclasses |
| 358 | and slot definitions for the class @code{<my-class>}. (@code{#:supers} |
| 359 | and @code{#:slots} are initialization keywords for the @code{dsupers} |
| 360 | and @code{dslots} slots of the @code{<class>} class.) |
| 361 | |
| 362 | In order to take advantage of the full power of the GOOPS metaobject |
| 363 | protocol (@pxref{MOP Specification}), it is sometimes desirable to |
| 364 | create a new class with a metaclass other than the default |
| 365 | @code{<class>}. This is done by writing: |
| 366 | |
| 367 | @example |
| 368 | (define-class <my-class2> (<object>) |
| 369 | slot @dots{} |
| 370 | #:metaclass <my-metaclass>) |
| 371 | @end example |
| 372 | |
| 373 | GOOPS expands this to something like: |
| 374 | |
| 375 | @example |
| 376 | (define <my-class2> |
| 377 | (make <my-metaclass> #:supers (list <object>) #:slots slots)) |
| 378 | @end example |
| 379 | |
| 380 | In this case, the value of @code{<my-class2>} is an instance of the more |
| 381 | specialized class @code{<my-metaclass>}. Note that |
| 382 | @code{<my-metaclass>} itself must previously have been defined as a |
| 383 | subclass of @code{<class>}. For a full discussion of when and how it is |
| 384 | useful to define new metaclasses, see @ref{MOP Specification}. |
| 385 | |
| 386 | Now let's make an instance of @code{<my-class2>}: |
| 387 | |
| 388 | @example |
| 389 | (define my-object (make <my-class2> ...)) |
| 390 | @end example |
| 391 | |
| 392 | All of the following statements are correct expressions of the |
| 393 | relationships between @code{my-object}, @code{<my-class2>}, |
| 394 | @code{<my-metaclass>} and @code{<class>}. |
| 395 | |
| 396 | @itemize @bullet |
| 397 | @item |
| 398 | @code{my-object} is an instance of the class @code{<my-class2>}. |
| 399 | |
| 400 | @item |
| 401 | @code{<my-class2>} is an instance of the class @code{<my-metaclass>}. |
| 402 | |
| 403 | @item |
| 404 | @code{<my-metaclass>} is an instance of the class @code{<class>}. |
| 405 | |
| 406 | @item |
| 407 | The class of @code{my-object} is @code{<my-class2>}. |
| 408 | |
| 409 | @item |
| 410 | The metaclass of @code{my-object} is @code{<my-metaclass>}. |
| 411 | |
| 412 | @item |
| 413 | The class of @code{<my-class2>} is @code{<my-metaclass>}. |
| 414 | |
| 415 | @item |
| 416 | The metaclass of @code{<my-class2>} is @code{<class>}. |
| 417 | |
| 418 | @item |
| 419 | The class of @code{<my-metaclass>} is @code{<class>}. |
| 420 | |
| 421 | @item |
| 422 | The metaclass of @code{<my-metaclass>} is @code{<class>}. |
| 423 | |
| 424 | @item |
| 425 | @code{<my-class2>} is not a metaclass, since it is does not inherit from |
| 426 | @code{<class>}. |
| 427 | |
| 428 | @item |
| 429 | @code{<my-metaclass>} is a metaclass, since it inherits from |
| 430 | @code{<class>}. |
| 431 | @end itemize |
| 432 | |
| 433 | @subsubheading Class Precedence List |
| 434 | |
| 435 | The @dfn{class precedence list} of a class is the list of all direct and |
| 436 | indirect superclasses of that class, including the class itself. |
| 437 | |
| 438 | In the absence of multiple inheritance, the class precedence list is |
| 439 | ordered straightforwardly, beginning with the class itself and ending |
| 440 | with @code{<top>}. |
| 441 | |
| 442 | For example, given this inheritance hierarchy: |
| 443 | |
| 444 | @example |
| 445 | (define-class <invertebrate> (<object>) @dots{}) |
| 446 | (define-class <echinoderm> (<invertebrate>) @dots{}) |
| 447 | (define-class <starfish> (<echinoderm>) @dots{}) |
| 448 | @end example |
| 449 | |
| 450 | the class precedence list of <starfish> would be |
| 451 | |
| 452 | @example |
| 453 | (<starfish> <echinoderm> <invertebrate> <object> <top>) |
| 454 | @end example |
| 455 | |
| 456 | With multiple inheritance, the algorithm is a little more complicated. |
| 457 | A full description is provided by the GOOPS Tutorial: see @ref{Class |
| 458 | precedence list}. |
| 459 | |
| 460 | ``Class precedence list'' is often abbreviated, in documentation and |
| 461 | Scheme variable names, to @dfn{cpl}. |
| 462 | |
| 463 | @subsubheading Accessor |
| 464 | |
| 465 | An @dfn{accessor} is a generic function with both reference and setter |
| 466 | methods. |
| 467 | |
| 468 | @example |
| 469 | (define-accessor perimeter) |
| 470 | @end example |
| 471 | |
| 472 | Reference methods for an accessor are defined in the same way as generic |
| 473 | function methods. |
| 474 | |
| 475 | @example |
| 476 | (define-method (perimeter (s <square>)) |
| 477 | (* 4 (side-length s))) |
| 478 | @end example |
| 479 | |
| 480 | Setter methods for an accessor are defined by specifying ``(setter |
| 481 | <accessor-name>)'' as the first parameter of the @code{define-method} |
| 482 | call. |
| 483 | |
| 484 | @example |
| 485 | (define-method ((setter perimeter) (s <square>) (n <number>)) |
| 486 | (set! (side-length s) (/ n 4))) |
| 487 | @end example |
| 488 | |
| 489 | Once an appropriate setter method has been defined in this way, it can |
| 490 | be invoked using the generalized @code{set!} syntax, as in: |
| 491 | |
| 492 | @example |
| 493 | (set! (perimeter s1) 18.3) |
| 494 | @end example |
| 495 | |
| 496 | @node Defining New Classes |
| 497 | @subsection Defining New Classes |
| 498 | |
| 499 | [ *fixme* Somewhere in this manual there needs to be an introductory |
| 500 | discussion about GOOPS classes, generic functions and methods, covering |
| 501 | |
| 502 | @itemize @bullet |
| 503 | @item |
| 504 | how classes encapsulate related items of data in @dfn{slots} |
| 505 | |
| 506 | @item |
| 507 | why it is that, unlike in C++ and Java, a class does not encapsulate the |
| 508 | methods that act upon the class (at least not in the C++/Java sense) |
| 509 | |
| 510 | @item |
| 511 | how generic functions provide a more general solution that provides for |
| 512 | dispatch on all argument types, and avoids idiosyncracies like C++'s |
| 513 | friend classes |
| 514 | |
| 515 | @item |
| 516 | how encapsulation in the sense of data- and code-hiding, or of |
| 517 | distinguishing interface from implementation, is treated in Guile as an |
| 518 | orthogonal concept to object orientation, and is the responsibility of |
| 519 | the module system. |
| 520 | @end itemize |
| 521 | |
| 522 | Some of this is covered in the Tutorial chapter, in @ref{Generic |
| 523 | functions and methods} - perhaps the best solution would be to expand |
| 524 | the discussion there. ] |
| 525 | |
| 526 | @menu |
| 527 | * Basic Class Definition:: |
| 528 | * Class Options:: |
| 529 | * Slot Options:: |
| 530 | * Class Definition Internals:: |
| 531 | * Customizing Class Definition:: |
| 532 | * STKlos Compatibility:: |
| 533 | @end menu |
| 534 | |
| 535 | @node Basic Class Definition |
| 536 | @subsubsection Basic Class Definition |
| 537 | |
| 538 | New classes are defined using the @code{define-class} syntax, with |
| 539 | arguments that specify the classes that the new class should inherit |
| 540 | from, the direct slots of the new class, and any required class options. |
| 541 | |
| 542 | @deffn syntax define-class name (super @dots{}) slot-definition @dots{} . options |
| 543 | Define a class called @var{name} that inherits from @var{super}s, with |
| 544 | direct slots defined by @var{slot-definition}s and class options |
| 545 | @var{options}. The newly created class is bound to the variable name |
| 546 | @var{name} in the current environment. |
| 547 | |
| 548 | Each @var{slot-definition} is either a symbol that names the slot or a |
| 549 | list, |
| 550 | |
| 551 | @example |
| 552 | (@var{slot-name-symbol} . @var{slot-options}) |
| 553 | @end example |
| 554 | |
| 555 | where @var{slot-name-symbol} is a symbol and @var{slot-options} is a |
| 556 | list with an even number of elements. The even-numbered elements of |
| 557 | @var{slot-options} (counting from zero) are slot option keywords; the |
| 558 | odd-numbered elements are the corresponding values for those keywords. |
| 559 | |
| 560 | @var{options} is a similarly structured list containing class option |
| 561 | keywords and corresponding values. |
| 562 | @end deffn |
| 563 | |
| 564 | The standard GOOPS class and slot options are described in the following |
| 565 | subsubsections: see @ref{Class Options} and @ref{Slot Options}. |
| 566 | |
| 567 | Example 1. Define a class that combines two pre-existing classes by |
| 568 | inheritance but adds no new slots. |
| 569 | |
| 570 | @example |
| 571 | (define-class <combined> (<tree> <bicycle>)) |
| 572 | @end example |
| 573 | |
| 574 | Example 2. Define a @code{regular-polygon} class with slots for side |
| 575 | length and number of sides that have default values and can be accessed |
| 576 | via the generic functions @code{side-length} and @code{num-sides}. |
| 577 | |
| 578 | @example |
| 579 | (define-class <regular-polygon> () |
| 580 | (sl #:init-value 1 #:accessor side-length) |
| 581 | (ns #:init-value 5 #:accessor num-sides)) |
| 582 | @end example |
| 583 | |
| 584 | Example 3. Define a class whose behavior (and that of its instances) is |
| 585 | customized via an application-defined metaclass. |
| 586 | |
| 587 | @example |
| 588 | (define-class <tcpip-fsm> () |
| 589 | (s #:init-value #f #:accessor state) |
| 590 | ... |
| 591 | #:metaclass <finite-state-class>) |
| 592 | @end example |
| 593 | |
| 594 | @node Class Options |
| 595 | @subsubsection Class Options |
| 596 | |
| 597 | @deffn {class option} #:metaclass metaclass |
| 598 | The @code{#:metaclass} class option specifies the metaclass of the class |
| 599 | being defined. @var{metaclass} must be a class that inherits from |
| 600 | @code{<class>}. For an introduction to the use of metaclasses, see |
| 601 | @ref{Metaobjects and the Metaobject Protocol} and @ref{Terminology}. |
| 602 | |
| 603 | If the @code{#:metaclass} option is absent, GOOPS reuses or constructs a |
| 604 | metaclass for the new class by calling @code{ensure-metaclass} |
| 605 | (@pxref{Class Definition Internals,, ensure-metaclass}). |
| 606 | @end deffn |
| 607 | |
| 608 | @deffn {class option} #:name name |
| 609 | The @code{#:name} class option specifies the new class's name. This |
| 610 | name is used to identify the class whenever related objects - the class |
| 611 | itself, its instances and its subclasses - are printed. |
| 612 | |
| 613 | If the @code{#:name} option is absent, GOOPS uses the first argument to |
| 614 | @code{define-class} as the class name. |
| 615 | @end deffn |
| 616 | |
| 617 | @deffn {class option} #:environment environment |
| 618 | *fixme* Not sure about this one, but I think that the |
| 619 | @code{#:environment} option specifies the environment in which the |
| 620 | class's getters and setters are computed and evaluated. |
| 621 | |
| 622 | If the @code{#:environment} option is not specified, the class's |
| 623 | environment defaults to the top-level environment in which the |
| 624 | @code{define-class} form appears. |
| 625 | @end deffn |
| 626 | |
| 627 | @node Slot Options |
| 628 | @subsubsection Slot Options |
| 629 | |
| 630 | @deffn {slot option} #:allocation allocation |
| 631 | The @code{#:allocation} option tells GOOPS how to allocate storage for |
| 632 | the slot. Possible values for @var{allocation} are |
| 633 | |
| 634 | @itemize @bullet |
| 635 | @item @code{#:instance} |
| 636 | |
| 637 | Indicates that GOOPS should create separate storage for this slot in |
| 638 | each new instance of the containing class (and its subclasses). |
| 639 | |
| 640 | @item @code{#:class} |
| 641 | |
| 642 | Indicates that GOOPS should create storage for this slot that is shared |
| 643 | by all instances of the containing class (and its subclasses). In other |
| 644 | words, a slot in class @var{C} with allocation @code{#:class} is shared |
| 645 | by all @var{instance}s for which @code{(is-a? @var{instance} @var{c})}. |
| 646 | |
| 647 | @item @code{#:each-subclass} |
| 648 | |
| 649 | Indicates that GOOPS should create storage for this slot that is shared |
| 650 | by all @emph{direct} instances of the containing class, and that |
| 651 | whenever a subclass of the containing class is defined, GOOPS should |
| 652 | create a new storage for the slot that is shared by all @emph{direct} |
| 653 | instances of the subclass. In other words, a slot with allocation |
| 654 | @code{#:each-subclass} is shared by all instances with the same |
| 655 | @code{class-of}. |
| 656 | |
| 657 | @item @code{#:virtual} |
| 658 | |
| 659 | Indicates that GOOPS should not allocate storage for this slot. The |
| 660 | slot definition must also include the @code{#:slot-ref} and |
| 661 | @code{#:slot-set!} options to specify how to reference and set the value |
| 662 | for this slot. |
| 663 | @end itemize |
| 664 | |
| 665 | The default value is @code{#:instance}. |
| 666 | |
| 667 | Slot allocation options are processed when defining a new class by the |
| 668 | generic function @code{compute-get-n-set}, which is specialized by the |
| 669 | class's metaclass. Hence new types of slot allocation can be |
| 670 | implemented by defining a new metaclass and a method for |
| 671 | @code{compute-get-n-set} that is specialized for the new metaclass. For |
| 672 | an example of how to do this, see @ref{Customizing Class Definition}. |
| 673 | @end deffn |
| 674 | |
| 675 | @deffn {slot option} #:slot-ref getter |
| 676 | @deffnx {slot option} #:slot-set! setter |
| 677 | The @code{#:slot-ref} and @code{#:slot-set!} options must be specified |
| 678 | if the slot allocation is @code{#:virtual}, and are ignored otherwise. |
| 679 | |
| 680 | @var{getter} should be a closure taking a single @var{instance} parameter |
| 681 | that returns the current slot value. @var{setter} should be a closure |
| 682 | taking two parameters - @var{instance} and @var{new-val} - that sets the |
| 683 | slot value to @var{new-val}. |
| 684 | @end deffn |
| 685 | |
| 686 | @deffn {slot option} #:getter getter |
| 687 | @deffnx {slot option} #:setter setter |
| 688 | @deffnx {slot option} #:accessor accessor |
| 689 | These options, if present, tell GOOPS to create generic function and |
| 690 | method definitions that can be used to get and set the slot value more |
| 691 | conveniently than by using @code{slot-ref} and @code{slot-set!}. |
| 692 | |
| 693 | @var{getter} specifies a generic function to which GOOPS will add a |
| 694 | method for getting the slot value. @var{setter} specifies a generic |
| 695 | function to which GOOPS will add a method for setting the slot value. |
| 696 | @var{accessor} specifies an accessor to which GOOPS will add methods for |
| 697 | both getting and setting the slot value. |
| 698 | |
| 699 | So if a class includes a slot definition like this: |
| 700 | |
| 701 | @example |
| 702 | (c #:getter get-count #:setter set-count #:accessor count) |
| 703 | @end example |
| 704 | |
| 705 | GOOPS defines generic function methods such that the slot value can be |
| 706 | referenced using either the getter or the accessor - |
| 707 | |
| 708 | @example |
| 709 | (let ((current-count (get-count obj))) @dots{}) |
| 710 | (let ((current-count (count obj))) @dots{}) |
| 711 | @end example |
| 712 | |
| 713 | - and set using either the setter or the accessor - |
| 714 | |
| 715 | @example |
| 716 | (set-count obj (+ 1 current-count)) |
| 717 | (set! (count obj) (+ 1 current-count)) |
| 718 | @end example |
| 719 | |
| 720 | Note that |
| 721 | |
| 722 | @itemize @bullet |
| 723 | @item |
| 724 | with an accessor, the slot value is set using the generalized |
| 725 | @code{set!} syntax |
| 726 | |
| 727 | @item |
| 728 | in practice, it is unusual for a slot to use all three of these options: |
| 729 | read-only, write-only and read-write slots would typically use only |
| 730 | @code{#:getter}, @code{#:setter} and @code{#:accessor} options |
| 731 | respectively. |
| 732 | @end itemize |
| 733 | |
| 734 | If the specified names are already bound in the top-level environment to |
| 735 | values that cannot be upgraded to generic functions, those values are |
| 736 | overwritten during evaluation of the @code{define-class} that contains |
| 737 | the slot definition. For details, see @ref{Generic Function Internals,, |
| 738 | ensure-generic}. |
| 739 | @end deffn |
| 740 | |
| 741 | @deffn {slot option} #:init-value init-value |
| 742 | @deffnx {slot option} #:init-form init-form |
| 743 | @deffnx {slot option} #:init-thunk init-thunk |
| 744 | @deffnx {slot option} #:init-keyword init-keyword |
| 745 | These options provide various ways to specify how to initialize the |
| 746 | slot's value at instance creation time. @var{init-value} is a fixed |
| 747 | value (shared across all new instances of the class). |
| 748 | @var{init-thunk} is a procedure of no arguments that is called |
| 749 | when a new instance is created and should return the desired initial |
| 750 | slot value. @var{init-form} is an unevaluated expression that gets |
| 751 | evaluated when a new instance is created and should return the desired |
| 752 | initial slot value. @var{init-keyword} is a keyword that can be used |
| 753 | to pass an initial slot value to @code{make} when creating a new |
| 754 | instance. |
| 755 | |
| 756 | Note that, since an @code{init-value} value is shared across all |
| 757 | instances of a class, you should only use it when the initial value is |
| 758 | an immutable value, like a constant. If you want to initialize a slot |
| 759 | with a fresh, independently mutable value, you should use |
| 760 | @code{init-thunk} or @code{init-form} instead. Consider the following |
| 761 | example. |
| 762 | |
| 763 | @example |
| 764 | (define-class <chbouib> () |
| 765 | (hashtab #:init-value (make-hash-table))) |
| 766 | @end example |
| 767 | |
| 768 | @noindent |
| 769 | Here only one hash table is created and all instances of |
| 770 | @code{<chbouib>} have their @code{hashtab} slot refer to it. In order |
| 771 | to have each instance of @code{<chbouib>} refer to a new hash table, you |
| 772 | should instead write: |
| 773 | |
| 774 | @example |
| 775 | (define-class <chbouib> () |
| 776 | (hashtab #:init-thunk make-hash-table)) |
| 777 | @end example |
| 778 | |
| 779 | @noindent |
| 780 | or: |
| 781 | |
| 782 | @example |
| 783 | (define-class <chbouib> () |
| 784 | (hashtab #:init-form (make-hash-table))) |
| 785 | @end example |
| 786 | |
| 787 | If more than one of these options is specified for the same slot, the |
| 788 | order of precedence, highest first is |
| 789 | |
| 790 | @itemize @bullet |
| 791 | @item |
| 792 | @code{#:init-keyword}, if @var{init-keyword} is present in the options |
| 793 | passed to @code{make} |
| 794 | |
| 795 | @item |
| 796 | @code{#:init-thunk}, @code{#:init-form} or @code{#:init-value}. |
| 797 | @end itemize |
| 798 | |
| 799 | If the slot definition contains more than one initialization option of |
| 800 | the same precedence, the later ones are ignored. If a slot is not |
| 801 | initialized at all, its value is unbound. |
| 802 | |
| 803 | In general, slots that are shared between more than one instance are |
| 804 | only initialized at new instance creation time if the slot value is |
| 805 | unbound at that time. However, if the new instance creation specifies |
| 806 | a valid init keyword and value for a shared slot, the slot is |
| 807 | re-initialized regardless of its previous value. |
| 808 | |
| 809 | Note, however, that the power of GOOPS' metaobject protocol means that |
| 810 | everything written here may be customized or overridden for particular |
| 811 | classes! The slot initializations described here are performed by the least |
| 812 | specialized method of the generic function @code{initialize}, whose |
| 813 | signature is |
| 814 | |
| 815 | @example |
| 816 | (define-method (initialize (object <object>) initargs) ...) |
| 817 | @end example |
| 818 | |
| 819 | The initialization of instances of any given class can be customized by |
| 820 | defining a @code{initialize} method that is specialized for that class, |
| 821 | and the author of the specialized method may decide to call |
| 822 | @code{next-method} - which will result in a call to the next less |
| 823 | specialized @code{initialize} method - at any point within the |
| 824 | specialized code, or maybe not at all. In general, therefore, the |
| 825 | initialization mechanisms described here may be modified or overridden by |
| 826 | more specialized code, or may not be supported at all for particular |
| 827 | classes. |
| 828 | @end deffn |
| 829 | |
| 830 | @node Class Definition Internals |
| 831 | @subsubsection Class Definition Internals |
| 832 | |
| 833 | Implementation notes: @code{define-class} expands to an expression which |
| 834 | |
| 835 | @itemize @bullet |
| 836 | @item |
| 837 | checks that it is being evaluated only at top level |
| 838 | |
| 839 | @item |
| 840 | defines any accessors that are implied by the @var{slot-definition}s |
| 841 | |
| 842 | @item |
| 843 | uses @code{class} to create the new class (@pxref{Class Definition |
| 844 | Internals,, class}) |
| 845 | |
| 846 | @item |
| 847 | checks for a previous class definition for @var{name} and, if found, |
| 848 | handles the redefinition by invoking @code{class-redefinition} |
| 849 | (@pxref{Redefining a Class}). |
| 850 | @end itemize |
| 851 | |
| 852 | @deffn syntax class name (super @dots{}) slot-definition @dots{} . options |
| 853 | Return a newly created class that inherits from @var{super}s, with |
| 854 | direct slots defined by @var{slot-definition}s and class options |
| 855 | @var{options}. For the format of @var{slot-definition}s and |
| 856 | @var{options}, see @ref{Basic Class Definition,, define-class}. |
| 857 | @end deffn |
| 858 | |
| 859 | Implementation notes: @code{class} expands to an expression which |
| 860 | |
| 861 | @itemize @bullet |
| 862 | @item |
| 863 | processes the class and slot definition options to check that they are |
| 864 | well-formed, to convert the @code{#:init-form} option to an |
| 865 | @code{#:init-thunk} option, to supply a default environment parameter |
| 866 | (the current top-level environment) and to evaluate all the bits that |
| 867 | need to be evaluated |
| 868 | |
| 869 | @item |
| 870 | calls @code{make-class} to create the class with the processed and |
| 871 | evaluated parameters. |
| 872 | @end itemize |
| 873 | |
| 874 | @deffn procedure make-class supers slots . options |
| 875 | Return a newly created class that inherits from @var{supers}, with |
| 876 | direct slots defined by @var{slots} and class options @var{options}. |
| 877 | For the format of @var{slots} and @var{options}, see @ref{Basic Class |
| 878 | Definition,, define-class}, except note that for @code{make-class}, |
| 879 | @var{slots} and @var{options} are separate list parameters: @var{slots} |
| 880 | here is a list of slot definitions. |
| 881 | @end deffn |
| 882 | |
| 883 | Implementation notes: @code{make-class} |
| 884 | |
| 885 | @itemize @bullet |
| 886 | @item |
| 887 | adds @code{<object>} to the @var{supers} list if @var{supers} is empty |
| 888 | or if none of the classes in @var{supers} have @code{<object>} in their |
| 889 | class precedence list |
| 890 | |
| 891 | @item |
| 892 | defaults the @code{#:environment}, @code{#:name} and @code{#:metaclass} |
| 893 | options, if they are not specified by @var{options}, to the current |
| 894 | top-level environment, the unbound value, and @code{(ensure-metaclass |
| 895 | @var{supers})} respectively (@pxref{Class Definition Internals,, |
| 896 | ensure-metaclass}) |
| 897 | |
| 898 | @item |
| 899 | checks for duplicate classes in @var{supers} and duplicate slot names in |
| 900 | @var{slots}, and signals an error if there are any duplicates |
| 901 | |
| 902 | @item |
| 903 | calls @code{make}, passing the metaclass as the first parameter and all |
| 904 | other parameters as option keywords with values. |
| 905 | @end itemize |
| 906 | |
| 907 | @deffn procedure ensure-metaclass supers env |
| 908 | Return a metaclass suitable for a class that inherits from the list of |
| 909 | classes in @var{supers}. The returned metaclass is the union by |
| 910 | inheritance of the metaclasses of the classes in @var{supers}. |
| 911 | |
| 912 | In the simplest case, where all the @var{supers} are straightforward |
| 913 | classes with metaclass @code{<class>}, the returned metaclass is just |
| 914 | @code{<class>}. |
| 915 | |
| 916 | For a more complex example, suppose that @var{supers} contained one |
| 917 | class with metaclass @code{<operator-class>} and one with metaclass |
| 918 | @code{<foreign-object-class>}. Then the returned metaclass would be a |
| 919 | class that inherits from both @code{<operator-class>} and |
| 920 | @code{<foreign-object-class>}. |
| 921 | |
| 922 | If @var{supers} is the empty list, @code{ensure-metaclass} returns the |
| 923 | default GOOPS metaclass @code{<class>}. |
| 924 | |
| 925 | GOOPS keeps a list of the metaclasses created by |
| 926 | @code{ensure-metaclass}, so that each required type of metaclass only |
| 927 | has to be created once. |
| 928 | |
| 929 | The @code{env} parameter is ignored. |
| 930 | @end deffn |
| 931 | |
| 932 | @deffn procedure ensure-metaclass-with-supers meta-supers |
| 933 | @code{ensure-metaclass-with-supers} is an internal procedure used by |
| 934 | @code{ensure-metaclass} (@pxref{Class Definition Internals,, |
| 935 | ensure-metaclass}). It returns a metaclass that is the union by |
| 936 | inheritance of the metaclasses in @var{meta-supers}. |
| 937 | @end deffn |
| 938 | |
| 939 | The internals of @code{make}, which is ultimately used to create the new |
| 940 | class object, are described in @ref{Customizing Instance Creation}, |
| 941 | which covers the creation and initialization of instances in general. |
| 942 | |
| 943 | @node Customizing Class Definition |
| 944 | @subsubsection Customizing Class Definition |
| 945 | |
| 946 | During the initialization of a new class, GOOPS calls a number of generic |
| 947 | functions with the newly allocated class instance as the first |
| 948 | argument. Specifically, GOOPS calls the generic function |
| 949 | |
| 950 | @itemize @bullet |
| 951 | @item |
| 952 | (initialize @var{class} @dots{}) |
| 953 | @end itemize |
| 954 | |
| 955 | where @var{class} is the newly allocated class instance, and the default |
| 956 | @code{initialize} method for arguments of type @code{<class>} calls the |
| 957 | generic functions |
| 958 | |
| 959 | @itemize @bullet |
| 960 | @item |
| 961 | (compute-cpl @var{class}) |
| 962 | |
| 963 | @item |
| 964 | (compute-slots @var{class}) |
| 965 | |
| 966 | @item |
| 967 | (compute-get-n-set @var{class} @var{slot-def}), for each of the slot |
| 968 | definitions returned by @code{compute-slots} |
| 969 | |
| 970 | @item |
| 971 | (compute-getter-method @var{class} @var{slot-def}), for each of the |
| 972 | slot definitions returned by @code{compute-slots} that includes a |
| 973 | @code{#:getter} or @code{#:accessor} slot option |
| 974 | |
| 975 | @item |
| 976 | (compute-setter-method @var{class} @var{slot-def}), for each of the |
| 977 | slot definitions returned by @code{compute-slots} that includes a |
| 978 | @code{#:setter} or @code{#:accessor} slot option. |
| 979 | @end itemize |
| 980 | |
| 981 | If the metaclass of the new class is something more specialized than the |
| 982 | default @code{<class>}, then the type of @var{class} in the calls above |
| 983 | is more specialized than @code{<class>}, and hence it becomes possible |
| 984 | to define generic function methods, specialized for the new class's |
| 985 | metaclass, that can modify or override the default behaviour of |
| 986 | @code{initialize}, @code{compute-cpl} or @code{compute-get-n-set}. |
| 987 | |
| 988 | @code{compute-cpl} computes the class precedence list (``CPL'') for the |
| 989 | new class (@pxref{Class precedence list}), and returns it as a list of |
| 990 | class objects. The CPL is important because it defines a superclass |
| 991 | ordering that is used, when a generic function is invoked upon an |
| 992 | instance of the class, to decide which of the available generic function |
| 993 | methods is the most specific. Hence @code{compute-cpl} could be |
| 994 | customized in order to modify the CPL ordering algorithm for all classes |
| 995 | with a special metaclass. |
| 996 | |
| 997 | The default CPL algorithm is encapsulated by the @code{compute-std-cpl} |
| 998 | procedure, which is in turn called by the default @code{compute-cpl} |
| 999 | method. |
| 1000 | |
| 1001 | @deffn procedure compute-std-cpl class |
| 1002 | Compute and return the class precedence list for @var{class} according |
| 1003 | to the algorithm described in @ref{Class precedence list}. |
| 1004 | @end deffn |
| 1005 | |
| 1006 | @code{compute-slots} computes and returns a list of all slot definitions |
| 1007 | for the new class. By default, this list includes the direct slot |
| 1008 | definitions from the @code{define-class} form, plus the slot definitions |
| 1009 | that are inherited from the new class's superclasses. The default |
| 1010 | @code{compute-slots} method uses the CPL computed by @code{compute-cpl} |
| 1011 | to calculate this union of slot definitions, with the rule that slots |
| 1012 | inherited from superclasses are shadowed by direct slots with the same |
| 1013 | name. One possible reason for customizing @code{compute-slots} would be |
| 1014 | to implement an alternative resolution strategy for slot name conflicts. |
| 1015 | |
| 1016 | @code{compute-get-n-set} computes the low-level closures that will be |
| 1017 | used to get and set the value of a particular slot, and returns them in |
| 1018 | a list with two elements. |
| 1019 | |
| 1020 | The closures returned depend on how storage for that slot is allocated. |
| 1021 | The standard @code{compute-get-n-set} method, specialized for classes of |
| 1022 | type @code{<class>}, handles the standard GOOPS values for the |
| 1023 | @code{#:allocation} slot option (@pxref{Slot Options,, allocation}). By |
| 1024 | defining a new @code{compute-get-n-set} method for a more specialized |
| 1025 | metaclass, it is possible to support new types of slot allocation. |
| 1026 | |
| 1027 | Suppose you wanted to create a large number of instances of some class |
| 1028 | with a slot that should be shared between some but not all instances of |
| 1029 | that class - say every 10 instances should share the same slot storage. |
| 1030 | The following example shows how to implement and use a new type of slot |
| 1031 | allocation to do this. |
| 1032 | |
| 1033 | @example |
| 1034 | (define-class <batched-allocation-metaclass> (<class>)) |
| 1035 | |
| 1036 | (let ((batch-allocation-count 0) |
| 1037 | (batch-get-n-set #f)) |
| 1038 | (define-method (compute-get-n-set (class <batched-allocation-metaclass>) s) |
| 1039 | (case (slot-definition-allocation s) |
| 1040 | ((#:batched) |
| 1041 | ;; If we've already used the same slot storage for 10 instances, |
| 1042 | ;; reset variables. |
| 1043 | (if (= batch-allocation-count 10) |
| 1044 | (begin |
| 1045 | (set! batch-allocation-count 0) |
| 1046 | (set! batch-get-n-set #f))) |
| 1047 | ;; If we don't have a current pair of get and set closures, |
| 1048 | ;; create one. make-closure-variable returns a pair of closures |
| 1049 | ;; around a single Scheme variable - see goops.scm for details. |
| 1050 | (or batch-get-n-set |
| 1051 | (set! batch-get-n-set (make-closure-variable))) |
| 1052 | ;; Increment the batch allocation count. |
| 1053 | (set! batch-allocation-count (+ batch-allocation-count 1)) |
| 1054 | batch-get-n-set) |
| 1055 | |
| 1056 | ;; Call next-method to handle standard allocation types. |
| 1057 | (else (next-method))))) |
| 1058 | |
| 1059 | (define-class <class-using-batched-slot> () |
| 1060 | ... |
| 1061 | (c #:allocation #:batched) |
| 1062 | ... |
| 1063 | #:metaclass <batched-allocation-metaclass>) |
| 1064 | @end example |
| 1065 | |
| 1066 | The usage of @code{compute-getter-method} and @code{compute-setter-method} |
| 1067 | is described in @ref{MOP Specification}. |
| 1068 | |
| 1069 | @code{compute-cpl} and @code{compute-get-n-set} are called by the |
| 1070 | standard @code{initialize} method for classes whose metaclass is |
| 1071 | @code{<class>}. But @code{initialize} itself can also be modified, by |
| 1072 | defining an @code{initialize} method specialized to the new class's |
| 1073 | metaclass. Such a method could complete override the standard |
| 1074 | behaviour, by not calling @code{(next-method)} at all, but more |
| 1075 | typically it would perform additional class initialization steps before |
| 1076 | and/or after calling @code{(next-method)} for the standard behaviour. |
| 1077 | |
| 1078 | @node STKlos Compatibility |
| 1079 | @subsubsection STKlos Compatibility |
| 1080 | |
| 1081 | If the STKlos compatibility module is loaded, @code{define-class} is |
| 1082 | overwritten by a STKlos-specific definition; the standard GOOPS |
| 1083 | definition of @code{define-class} remains available in |
| 1084 | @code{standard-define-class}. |
| 1085 | |
| 1086 | @deffn syntax standard-define-class name (super @dots{}) slot-definition @dots{} . options |
| 1087 | @code{standard-define-class} is equivalent to the standard GOOPS |
| 1088 | @code{define-class}. |
| 1089 | @end deffn |
| 1090 | |
| 1091 | @node Creating Instances |
| 1092 | @subsection Creating Instances |
| 1093 | |
| 1094 | @menu |
| 1095 | * Basic Instance Creation:: |
| 1096 | * Customizing Instance Creation:: |
| 1097 | @end menu |
| 1098 | |
| 1099 | @node Basic Instance Creation |
| 1100 | @subsubsection Basic Instance Creation |
| 1101 | |
| 1102 | To create a new instance of any GOOPS class, use the generic function |
| 1103 | @code{make} or @code{make-instance}, passing the required class and any |
| 1104 | appropriate instance initialization arguments as keyword and value |
| 1105 | pairs. Note that @code{make} and @code{make-instances} are aliases for |
| 1106 | each other - their behaviour is identical. |
| 1107 | |
| 1108 | @deffn generic make |
| 1109 | @deffnx method make (class <class>) . initargs |
| 1110 | Create and return a new instance of class @var{class}, initialized using |
| 1111 | @var{initargs}. |
| 1112 | |
| 1113 | In theory, @var{initargs} can have any structure that is understood by |
| 1114 | whatever methods get applied when the @code{initialize} generic function |
| 1115 | is applied to the newly allocated instance. |
| 1116 | |
| 1117 | In practice, specialized @code{initialize} methods would normally call |
| 1118 | @code{(next-method)}, and so eventually the standard GOOPS |
| 1119 | @code{initialize} methods are applied. These methods expect |
| 1120 | @var{initargs} to be a list with an even number of elements, where |
| 1121 | even-numbered elements (counting from zero) are keywords and |
| 1122 | odd-numbered elements are the corresponding values. |
| 1123 | |
| 1124 | GOOPS processes initialization argument keywords automatically for slots |
| 1125 | whose definition includes the @code{#:init-keyword} option (@pxref{Slot |
| 1126 | Options,, init-keyword}). Other keyword value pairs can only be |
| 1127 | processed by an @code{initialize} method that is specialized for the new |
| 1128 | instance's class. Any unprocessed keyword value pairs are ignored. |
| 1129 | @end deffn |
| 1130 | |
| 1131 | @deffn generic make-instance |
| 1132 | @deffnx method make-instance (class <class>) . initargs |
| 1133 | @code{make-instance} is an alias for @code{make}. |
| 1134 | @end deffn |
| 1135 | |
| 1136 | @node Customizing Instance Creation |
| 1137 | @subsubsection Customizing Instance Creation |
| 1138 | |
| 1139 | @code{make} itself is a generic function. Hence the @code{make} |
| 1140 | invocation itself can be customized in the case where the new instance's |
| 1141 | metaclass is more specialized than the default @code{<class>}, by |
| 1142 | defining a @code{make} method that is specialized to that metaclass. |
| 1143 | |
| 1144 | Normally, however, the method for classes with metaclass @code{<class>} |
| 1145 | will be applied. This method calls two generic functions: |
| 1146 | |
| 1147 | @itemize @bullet |
| 1148 | @item |
| 1149 | (allocate-instance @var{class} . @var{initargs}) |
| 1150 | |
| 1151 | @item |
| 1152 | (initialize @var{instance} . @var{initargs}) |
| 1153 | @end itemize |
| 1154 | |
| 1155 | @code{allocate-instance} allocates storage for and returns the new |
| 1156 | instance, uninitialized. You might customize @code{allocate-instance}, |
| 1157 | for example, if you wanted to provide a GOOPS wrapper around some other |
| 1158 | object programming system. |
| 1159 | |
| 1160 | To do this, you would create a specialized metaclass, which would act as |
| 1161 | the metaclass for all classes and instances from the other system. Then |
| 1162 | define an @code{allocate-instance} method, specialized to that |
| 1163 | metaclass, which calls a Guile primitive C function, which in turn |
| 1164 | allocates the new instance using the interface of the other object |
| 1165 | system. |
| 1166 | |
| 1167 | In this case, for a complete system, you would also need to customize a |
| 1168 | number of other generic functions like @code{make} and |
| 1169 | @code{initialize}, so that GOOPS knows how to make classes from the |
| 1170 | other system, access instance slots, and so on. |
| 1171 | |
| 1172 | @code{initialize} initializes the instance that is returned by |
| 1173 | @code{allocate-instance}. The standard GOOPS methods perform |
| 1174 | initializations appropriate to the instance class. |
| 1175 | |
| 1176 | @itemize @bullet |
| 1177 | @item |
| 1178 | At the least specialized level, the method for instances of type |
| 1179 | @code{<object>} performs internal GOOPS instance initialization, and |
| 1180 | initializes the instance's slots according to the slot definitions and |
| 1181 | any slot initialization keywords that appear in @var{initargs}. |
| 1182 | |
| 1183 | @item |
| 1184 | The method for instances of type @code{<class>} calls |
| 1185 | @code{(next-method)}, then performs the class initializations described |
| 1186 | in @ref{Customizing Class Definition}. |
| 1187 | |
| 1188 | @item |
| 1189 | and so on for generic functions, method, operator classes @dots{} |
| 1190 | @end itemize |
| 1191 | |
| 1192 | Similarly, you can customize the initialization of instances of any |
| 1193 | application-defined class by defining an @code{initialize} method |
| 1194 | specialized to that class. |
| 1195 | |
| 1196 | Imagine a class whose instances' slots need to be initialized at |
| 1197 | instance creation time by querying a database. Although it might be |
| 1198 | possible to achieve this a combination of @code{#:init-thunk} keywords |
| 1199 | and closures in the slot definitions, it is neater to write an |
| 1200 | @code{initialize} method for the class that queries the database once |
| 1201 | and initializes all the dependent slot values according to the results. |
| 1202 | |
| 1203 | @node Accessing Slots |
| 1204 | @subsection Accessing Slots |
| 1205 | |
| 1206 | The definition of a slot contains at the very least a slot name, and may |
| 1207 | also contain various slot options, including getter, setter and/or |
| 1208 | accessor functions for the slot. |
| 1209 | |
| 1210 | It is always possible to access slots by name, using the various |
| 1211 | ``slot-ref'' and ``slot-set!'' procedures described in the following |
| 1212 | subsubsections. For example, |
| 1213 | |
| 1214 | @example |
| 1215 | (define-class <my-class> () ;; Define a class with slots |
| 1216 | (count #:init-value 0) ;; named "count" and "cache". |
| 1217 | (cache #:init-value '()) |
| 1218 | @dots{}) |
| 1219 | |
| 1220 | (define inst (make <my-class>)) ;; Make an instance of this class. |
| 1221 | |
| 1222 | (slot-set! inst 'count 5) ;; Set the value of the "count" |
| 1223 | ;; slot to 5. |
| 1224 | |
| 1225 | (slot-set! inst 'cache ;; Modify the value of the |
| 1226 | (cons (cons "^it" "It") ;; "cache" slot. |
| 1227 | (slot-ref inst 'cache))) |
| 1228 | @end example |
| 1229 | |
| 1230 | If a slot definition includes a getter, setter or accessor function, |
| 1231 | these can be used instead of @code{slot-ref} and @code{slot-set!} to |
| 1232 | access the slot. |
| 1233 | |
| 1234 | @example |
| 1235 | (define-class <adv-class> () ;; Define a new class whose slots |
| 1236 | (count #:setter set-count) ;; use a getter, a setter and |
| 1237 | (cache #:accessor cache) ;; an accessor. |
| 1238 | (csize #:getter cache-size) |
| 1239 | @dots{}) |
| 1240 | |
| 1241 | (define inst (make <adv-class>)) ;; Make an instance of this class. |
| 1242 | |
| 1243 | (set-count inst 5) ;; Set the value of the "count" |
| 1244 | ;; slot to 5. |
| 1245 | |
| 1246 | (set! (cache inst) ;; Modify the value of the |
| 1247 | (cons (cons "^it" "It") ;; "cache" slot. |
| 1248 | (cache inst))) |
| 1249 | |
| 1250 | (let ((size (cache-size inst))) ;; Get the value of the "csize" |
| 1251 | @dots{}) ;; slot. |
| 1252 | @end example |
| 1253 | |
| 1254 | Whichever of these methods is used to access slots, GOOPS always calls |
| 1255 | the low-level @dfn{getter} and @dfn{setter} closures for the slot to get |
| 1256 | and set its value. These closures make sure that the slot behaves |
| 1257 | according to the @code{#:allocation} type that was specified in the slot |
| 1258 | definition (@pxref{Slot Options,, allocation}). (For more about these |
| 1259 | closures, see @ref{Customizing Class Definition,, compute-get-n-set}.) |
| 1260 | |
| 1261 | @menu |
| 1262 | * Instance Slots:: |
| 1263 | * Class Slots:: |
| 1264 | * Handling Slot Access Errors:: |
| 1265 | @end menu |
| 1266 | |
| 1267 | @node Instance Slots |
| 1268 | @subsubsection Instance Slots |
| 1269 | |
| 1270 | Any slot, regardless of its allocation, can be queried, referenced and |
| 1271 | set using the following four primitive procedures. |
| 1272 | |
| 1273 | @deffn {primitive procedure} slot-exists? obj slot-name |
| 1274 | Return @code{#t} if @var{obj} has a slot with name @var{slot-name}, |
| 1275 | otherwise @code{#f}. |
| 1276 | @end deffn |
| 1277 | |
| 1278 | @deffn {primitive procedure} slot-bound? obj slot-name |
| 1279 | Return @code{#t} if the slot named @var{slot-name} in @var{obj} has a |
| 1280 | value, otherwise @code{#f}. |
| 1281 | |
| 1282 | @code{slot-bound?} calls the generic function @code{slot-missing} if |
| 1283 | @var{obj} does not have a slot called @var{slot-name} (@pxref{Handling |
| 1284 | Slot Access Errors, slot-missing}). |
| 1285 | @end deffn |
| 1286 | |
| 1287 | @deffn {primitive procedure} slot-ref obj slot-name |
| 1288 | Return the value of the slot named @var{slot-name} in @var{obj}. |
| 1289 | |
| 1290 | @code{slot-ref} calls the generic function @code{slot-missing} if |
| 1291 | @var{obj} does not have a slot called @var{slot-name} (@pxref{Handling |
| 1292 | Slot Access Errors, slot-missing}). |
| 1293 | |
| 1294 | @code{slot-ref} calls the generic function @code{slot-unbound} if the |
| 1295 | named slot in @var{obj} does not have a value (@pxref{Handling Slot |
| 1296 | Access Errors, slot-unbound}). |
| 1297 | @end deffn |
| 1298 | |
| 1299 | @deffn {primitive procedure} slot-set! obj slot-name value |
| 1300 | Set the value of the slot named @var{slot-name} in @var{obj} to @var{value}. |
| 1301 | |
| 1302 | @code{slot-set!} calls the generic function @code{slot-missing} if |
| 1303 | @var{obj} does not have a slot called @var{slot-name} (@pxref{Handling |
| 1304 | Slot Access Errors, slot-missing}). |
| 1305 | @end deffn |
| 1306 | |
| 1307 | GOOPS stores information about slots in class metaobjects. Internally, |
| 1308 | all of these procedures work by looking up the slot definition for the |
| 1309 | slot named @var{slot-name} in the class metaobject for @code{(class-of |
| 1310 | @var{obj})}, and then using the slot definition's ``getter'' and |
| 1311 | ``setter'' closures to get and set the slot value. |
| 1312 | |
| 1313 | The next four procedures differ from the previous ones in that they take |
| 1314 | the class metaobject as an explicit argument, rather than assuming |
| 1315 | @code{(class-of @var{obj})}. Therefore they allow you to apply the |
| 1316 | ``getter'' and ``setter'' closures of a slot definition in one class to |
| 1317 | an instance of a different class. |
| 1318 | |
| 1319 | [ *fixme* I have no idea why this is useful! Perhaps when a slot in |
| 1320 | @code{(class-of @var{obj})} shadows a slot with the same name in one of |
| 1321 | its superclasses? There should be an enlightening example here. ] |
| 1322 | |
| 1323 | @deffn {primitive procedure} slot-exists-using-class? class obj slot-name |
| 1324 | Return @code{#t} if the class metaobject @var{class} has a slot |
| 1325 | definition for a slot with name @var{slot-name}, otherwise @code{#f}. |
| 1326 | @end deffn |
| 1327 | |
| 1328 | @deffn {primitive procedure} slot-bound-using-class? class obj slot-name |
| 1329 | Return @code{#t} if applying @code{slot-ref-using-class} to the same |
| 1330 | arguments would call the generic function @code{slot-unbound}, otherwise |
| 1331 | @code{#f}. |
| 1332 | |
| 1333 | @code{slot-bound-using-class?} calls the generic function |
| 1334 | @code{slot-missing} if @var{class} does not have a slot definition for a |
| 1335 | slot called @var{slot-name} (@pxref{Handling Slot Access Errors, |
| 1336 | slot-missing}). |
| 1337 | @end deffn |
| 1338 | |
| 1339 | @deffn {primitive procedure} slot-ref-using-class class obj slot-name |
| 1340 | Apply the ``getter'' closure for the slot named @var{slot-name} in |
| 1341 | @var{class} to @var{obj}, and return its result. |
| 1342 | |
| 1343 | @code{slot-ref-using-class} calls the generic function |
| 1344 | @code{slot-missing} if @var{class} does not have a slot definition for a |
| 1345 | slot called @var{slot-name} (@pxref{Handling Slot Access Errors, |
| 1346 | slot-missing}). |
| 1347 | |
| 1348 | @code{slot-ref-using-class} calls the generic function |
| 1349 | @code{slot-unbound} if the application of the ``getter'' closure to |
| 1350 | @var{obj} returns an unbound value (@pxref{Handling Slot Access Errors, |
| 1351 | slot-unbound}). |
| 1352 | @end deffn |
| 1353 | |
| 1354 | @deffn {primitive procedure} slot-set-using-class! class obj slot-name value |
| 1355 | Apply the ``setter'' closure for the slot named @var{slot-name} in |
| 1356 | @var{class} to @var{obj} and @var{value}. |
| 1357 | |
| 1358 | @code{slot-set-using-class!} calls the generic function |
| 1359 | @code{slot-missing} if @var{class} does not have a slot definition for a |
| 1360 | slot called @var{slot-name} (@pxref{Handling Slot Access Errors, |
| 1361 | slot-missing}). |
| 1362 | @end deffn |
| 1363 | |
| 1364 | @node Class Slots |
| 1365 | @subsubsection Class Slots |
| 1366 | |
| 1367 | Slots whose allocation is per-class rather than per-instance can be |
| 1368 | referenced and set without needing to specify any particular instance. |
| 1369 | |
| 1370 | @deffn procedure class-slot-ref class slot-name |
| 1371 | Return the value of the slot named @var{slot-name} in class @var{class}. |
| 1372 | The named slot must have @code{#:class} or @code{#:each-subclass} |
| 1373 | allocation (@pxref{Slot Options,, allocation}). |
| 1374 | |
| 1375 | If there is no such slot with @code{#:class} or @code{#:each-subclass} |
| 1376 | allocation, @code{class-slot-ref} calls the @code{slot-missing} generic |
| 1377 | function with arguments @var{class} and @var{slot-name}. Otherwise, if |
| 1378 | the slot value is unbound, @code{class-slot-ref} calls the |
| 1379 | @code{slot-missing} generic function, with the same arguments. |
| 1380 | @end deffn |
| 1381 | |
| 1382 | @deffn procedure class-slot-set! class slot-name value |
| 1383 | Set the value of the slot named @var{slot-name} in class @var{class} to |
| 1384 | @var{value}. The named slot must have @code{#:class} or |
| 1385 | @code{#:each-subclass} allocation (@pxref{Slot Options,, allocation}). |
| 1386 | |
| 1387 | If there is no such slot with @code{#:class} or @code{#:each-subclass} |
| 1388 | allocation, @code{class-slot-ref} calls the @code{slot-missing} generic |
| 1389 | function with arguments @var{class} and @var{slot-name}. |
| 1390 | @end deffn |
| 1391 | |
| 1392 | @node Handling Slot Access Errors |
| 1393 | @subsubsection Handling Slot Access Errors |
| 1394 | |
| 1395 | GOOPS calls one of the following generic functions when a ``slot-ref'' |
| 1396 | or ``slot-set!'' call specifies a non-existent slot name, or tries to |
| 1397 | reference a slot whose value is unbound. |
| 1398 | |
| 1399 | @deffn generic slot-missing |
| 1400 | @deffnx method slot-missing (class <class>) slot-name |
| 1401 | @deffnx method slot-missing (class <class>) (object <object>) slot-name |
| 1402 | @deffnx method slot-missing (class <class>) (object <object>) slot-name value |
| 1403 | When an application attempts to reference or set a class or instance |
| 1404 | slot by name, and the slot name is invalid for the specified @var{class} |
| 1405 | or @var{object}, GOOPS calls the @code{slot-missing} generic function. |
| 1406 | |
| 1407 | The default methods all call @code{goops-error} with an appropriate |
| 1408 | message. |
| 1409 | @end deffn |
| 1410 | |
| 1411 | @deffn generic slot-unbound |
| 1412 | @deffnx method slot-unbound (object <object>) |
| 1413 | @deffnx method slot-unbound (class <class>) slot-name |
| 1414 | @deffnx method slot-unbound (class <class>) (object <object>) slot-name |
| 1415 | When an application attempts to reference a class or instance slot, and |
| 1416 | the slot's value is unbound, GOOPS calls the @code{slot-unbound} generic |
| 1417 | function. |
| 1418 | |
| 1419 | The default methods all call @code{goops-error} with an appropriate |
| 1420 | message. |
| 1421 | @end deffn |
| 1422 | |
| 1423 | @node Creating Generic Functions |
| 1424 | @subsection Creating Generic Functions |
| 1425 | |
| 1426 | A generic function is a collection of methods, with rules for |
| 1427 | determining which of the methods should be applied for any given |
| 1428 | invocation of the generic function. |
| 1429 | |
| 1430 | GOOPS represents generic functions as metaobjects of the class |
| 1431 | @code{<generic>} (or one of its subclasses). |
| 1432 | |
| 1433 | @menu |
| 1434 | * Basic Generic Function Creation:: |
| 1435 | * Generic Function Internals:: |
| 1436 | * Extending Guiles Primitives:: |
| 1437 | @end menu |
| 1438 | |
| 1439 | @node Basic Generic Function Creation |
| 1440 | @subsubsection Basic Generic Function Creation |
| 1441 | |
| 1442 | The following forms may be used to bind a variable to a generic |
| 1443 | function. Depending on that variable's pre-existing value, the generic |
| 1444 | function may be created empty - with no methods - or it may contain |
| 1445 | methods that are inferred from the pre-existing value. |
| 1446 | |
| 1447 | It is not, in general, necessary to use @code{define-generic} or |
| 1448 | @code{define-accessor} before defining methods for the generic function |
| 1449 | using @code{define-method}, since @code{define-method} will |
| 1450 | automatically interpolate a @code{define-generic} call, or upgrade an |
| 1451 | existing generic to an accessor, if that is implied by the |
| 1452 | @code{define-method} call. Note in particular that, |
| 1453 | if the specified variable already has a @emph{generic function} value, |
| 1454 | @code{define-generic} and @code{define-accessor} will @emph{discard} it! |
| 1455 | Obviously it is application-dependent whether this is desirable or not. |
| 1456 | |
| 1457 | If, for example, you wanted to extend @code{+} for a class representing |
| 1458 | a new numerical type, you probably want to inherit any existing methods |
| 1459 | for @code{+} and so should not use @code{define-generic}. If, on the |
| 1460 | other hand, you do not want to risk inheriting methods whose behaviour |
| 1461 | might surprise you, you can use @code{define-generic} or |
| 1462 | @code{define-accessor} to wipe the slate clean. |
| 1463 | |
| 1464 | @deffn syntax define-generic symbol |
| 1465 | Create a generic function with name @var{symbol} and bind it to the |
| 1466 | variable @var{symbol}. |
| 1467 | |
| 1468 | If the variable @var{symbol} was previously bound to a Scheme procedure |
| 1469 | (or procedure-with-setter), the old procedure (and setter) is |
| 1470 | incorporated into the new generic function as its default procedure (and |
| 1471 | setter). Any other previous value that was bound to @var{symbol}, |
| 1472 | including an existing generic function, is overwritten by the new |
| 1473 | generic function. |
| 1474 | @end deffn |
| 1475 | |
| 1476 | @deffn syntax define-accessor symbol |
| 1477 | Create an accessor with name @var{symbol} and bind it to the variable |
| 1478 | @var{symbol}. |
| 1479 | |
| 1480 | If the variable @var{symbol} was previously bound to a Scheme procedure |
| 1481 | (or procedure-with-setter), the old procedure (and setter) is |
| 1482 | incorporated into the new accessor as its default procedure (and |
| 1483 | setter). Any other previous value that was bound to @var{symbol}, |
| 1484 | including an existing generic function or accessor, is overwritten by |
| 1485 | the new definition. |
| 1486 | @end deffn |
| 1487 | |
| 1488 | It is sometimes tempting to use GOOPS accessors with short names. For |
| 1489 | example, it is tempting to use the name @code{x} for the x-coordinate |
| 1490 | in vector packages. |
| 1491 | |
| 1492 | Assume that we work with a graphical package which needs to use two |
| 1493 | independent vector packages for 2D and 3D vectors respectively. If |
| 1494 | both packages export @code{x} we will encounter a name collision. |
| 1495 | |
| 1496 | This can be resolved automagically with the duplicates handler |
| 1497 | @code{merge-generics} which gives the module system license to merge |
| 1498 | all generic functions sharing a common name: |
| 1499 | |
| 1500 | @smalllisp |
| 1501 | (define-module (math 2D-vectors) |
| 1502 | :use-module (oop goops) |
| 1503 | :export (x y ...)) |
| 1504 | |
| 1505 | (define-module (math 3D-vectors) |
| 1506 | :use-module (oop goops) |
| 1507 | :export (x y z ...)) |
| 1508 | |
| 1509 | (define-module (my-module) |
| 1510 | :use-module (math 2D-vectors) |
| 1511 | :use-module (math 3D-vectors) |
| 1512 | :duplicates merge-generics) |
| 1513 | @end smalllisp |
| 1514 | |
| 1515 | The generic function @code{x} in @code{(my-module)} will now share |
| 1516 | methods with @code{x} in both imported modules. |
| 1517 | |
| 1518 | There will, in fact, now be three distinct generic functions named |
| 1519 | @code{x}: @code{x} in @code{(2D-vectors)}, @code{x} in |
| 1520 | @code{(3D-vectors)}, and @code{x} in @code{(my-module)}. The last |
| 1521 | function will be an @code{<extended-generic>}, extending the previous |
| 1522 | two functions. |
| 1523 | |
| 1524 | Let's call the imported generic functions the "ancestor functions". |
| 1525 | The generic function @code{x} in @code{(my-module)} is, in turn, a |
| 1526 | "descendant function" of the imported functions, extending its |
| 1527 | ancestors. |
| 1528 | |
| 1529 | For any generic function G, the applicable methods are selected from |
| 1530 | the union of the methods of the descendant functions, the methods of G |
| 1531 | itself and the methods of the ancestor functions. |
| 1532 | |
| 1533 | This, ancestor functions share methods with their descendants and vice |
| 1534 | versa. This implies that @code{x} in @code{(math 2D-vectors)} will |
| 1535 | share the methods of @code{x} in @code{(my-module)} and vice versa, |
| 1536 | while @code{x} in @code{(math 2D-vectors)} doesn't share the methods |
| 1537 | of @code{x} in @code{(math 3D-vectors)}, thus preserving modularity. |
| 1538 | |
| 1539 | Sharing is dynamic, so that adding new methods to a descendant implies |
| 1540 | adding it to the ancestor. |
| 1541 | |
| 1542 | If duplicates checking is desired in the above example, the following |
| 1543 | form of the @code{:duplicates} option can be used instead: |
| 1544 | |
| 1545 | @smalllisp |
| 1546 | :duplicates (merge-generics check) |
| 1547 | @end smalllisp |
| 1548 | |
| 1549 | @node Generic Function Internals |
| 1550 | @subsubsection Generic Function Internals |
| 1551 | |
| 1552 | @code{define-generic} calls @code{ensure-generic} to upgrade a |
| 1553 | pre-existing procedure value, or @code{make} with metaclass |
| 1554 | @code{<generic>} to create a new generic function. |
| 1555 | |
| 1556 | @code{define-accessor} calls @code{ensure-accessor} to upgrade a |
| 1557 | pre-existing procedure value, or @code{make-accessor} to create a new |
| 1558 | accessor. |
| 1559 | |
| 1560 | @deffn procedure ensure-generic old-definition [name] |
| 1561 | Return a generic function with name @var{name}, if possible by using or |
| 1562 | upgrading @var{old-definition}. If unspecified, @var{name} defaults to |
| 1563 | @code{#f}. |
| 1564 | |
| 1565 | If @var{old-definition} is already a generic function, it is returned |
| 1566 | unchanged. |
| 1567 | |
| 1568 | If @var{old-definition} is a Scheme procedure or procedure-with-setter, |
| 1569 | @code{ensure-generic} returns a new generic function that uses |
| 1570 | @var{old-definition} for its default procedure and setter. |
| 1571 | |
| 1572 | Otherwise @code{ensure-generic} returns a new generic function with no |
| 1573 | defaults and no methods. |
| 1574 | @end deffn |
| 1575 | |
| 1576 | @deffn procedure make-generic [name] |
| 1577 | Return a new generic function with name @code{(car @var{name})}. If |
| 1578 | unspecified, @var{name} defaults to @code{#f}. |
| 1579 | @end deffn |
| 1580 | |
| 1581 | @code{ensure-generic} calls @code{make} with metaclasses |
| 1582 | @code{<generic>} and @code{<generic-with-setter>}, depending on the |
| 1583 | previous value of the variable that it is trying to upgrade. |
| 1584 | |
| 1585 | @code{make-generic} is a simple wrapper for @code{make} with metaclass |
| 1586 | @code{<generic>}. |
| 1587 | |
| 1588 | @deffn procedure ensure-accessor proc [name] |
| 1589 | Return an accessor with name @var{name}, if possible by using or |
| 1590 | upgrading @var{proc}. If unspecified, @var{name} defaults to @code{#f}. |
| 1591 | |
| 1592 | If @var{proc} is already an accessor, it is returned unchanged. |
| 1593 | |
| 1594 | If @var{proc} is a Scheme procedure, procedure-with-setter or generic |
| 1595 | function, @code{ensure-accessor} returns an accessor that reuses the |
| 1596 | reusable elements of @var{proc}. |
| 1597 | |
| 1598 | Otherwise @code{ensure-accessor} returns a new accessor with no defaults |
| 1599 | and no methods. |
| 1600 | @end deffn |
| 1601 | |
| 1602 | @deffn procedure make-accessor [name] |
| 1603 | Return a new accessor with name @code{(car @var{name})}. If |
| 1604 | unspecified, @var{name} defaults to @code{#f}. |
| 1605 | @end deffn |
| 1606 | |
| 1607 | @code{ensure-accessor} calls @code{make} with |
| 1608 | metaclass @code{<generic-with-setter>}, as well as calls to |
| 1609 | @code{ensure-generic}, @code{make-accessor} and (tail recursively) |
| 1610 | @code{ensure-accessor}. |
| 1611 | |
| 1612 | @code{make-accessor} calls @code{make} twice, first |
| 1613 | with metaclass @code{<generic>} to create a generic function for the |
| 1614 | setter, then with metaclass @code{<generic-with-setter>} to create the |
| 1615 | accessor, passing the setter generic function as the value of the |
| 1616 | @code{#:setter} keyword. |
| 1617 | |
| 1618 | @node Extending Guiles Primitives |
| 1619 | @subsubsection Extending Guile's Primitives |
| 1620 | |
| 1621 | When GOOPS is loaded, many of Guile's primitive procedures can be |
| 1622 | extended by giving them a generic function definition that operates |
| 1623 | in conjunction with their normal C-coded implementation. For |
| 1624 | primitives that are extended in this way, the result from the user- |
| 1625 | or application-level point of view is that the extended primitive |
| 1626 | behaves exactly like a generic function with the C-coded implementation |
| 1627 | as its default method. |
| 1628 | |
| 1629 | The @code{generic-capability?} predicate should be used to determine |
| 1630 | whether a particular primitive is extensible in this way. |
| 1631 | |
| 1632 | @deffn {primitive procedure} generic-capability? primitive |
| 1633 | Return @code{#t} if @var{primitive} can be extended by giving it a |
| 1634 | generic function definition, otherwise @code{#f}. |
| 1635 | @end deffn |
| 1636 | |
| 1637 | Even when a primitive procedure is extensible like this, its generic |
| 1638 | function definition is not created until it is needed by a call to |
| 1639 | @code{define-method}, or until the application explicitly requests it |
| 1640 | by calling @code{enable-primitive-generic!}. |
| 1641 | |
| 1642 | @deffn {primitive procedure} enable-primitive-generic! primitive |
| 1643 | Force the creation of a generic function definition for |
| 1644 | @var{primitive}. |
| 1645 | @end deffn |
| 1646 | |
| 1647 | Once the generic function definition for a primitive has been created, |
| 1648 | it can be retrieved using @code{primitive-generic-generic}. |
| 1649 | |
| 1650 | @deffn {primitive procedure} primitive-generic-generic primitive |
| 1651 | Return the generic function definition of @var{primitive}. |
| 1652 | |
| 1653 | @code{primitive-generic-generic} raises an error if @var{primitive} |
| 1654 | is not a primitive with generic capability, or if its generic capability |
| 1655 | has not yet been enabled, whether implicitly (by @code{define-method}) |
| 1656 | or explicitly (by @code{enable-primitive-generic!}). |
| 1657 | @end deffn |
| 1658 | |
| 1659 | Note that the distinction between, on the one hand, primitives with |
| 1660 | additional generic function definitions and, on the other hand, generic |
| 1661 | functions with a default method, may disappear when GOOPS is fully |
| 1662 | integrated into the core of Guile. Consequently, the |
| 1663 | procedures described in this section may disappear as well. |
| 1664 | |
| 1665 | @node Adding Methods to Generic Functions |
| 1666 | @subsection Adding Methods to Generic Functions |
| 1667 | |
| 1668 | @menu |
| 1669 | * Basic Method Definition:: |
| 1670 | * Method Definition Internals:: |
| 1671 | @end menu |
| 1672 | |
| 1673 | @node Basic Method Definition |
| 1674 | @subsubsection Basic Method Definition |
| 1675 | |
| 1676 | To add a method to a generic function, use the @code{define-method} form. |
| 1677 | |
| 1678 | @deffn syntax define-method (generic parameter @dots{}) . body |
| 1679 | Define a method for the generic function or accessor @var{generic} with |
| 1680 | parameters @var{parameter}s and body @var{body}. |
| 1681 | |
| 1682 | @var{generic} is a generic function. If @var{generic} is a variable |
| 1683 | which is not yet bound to a generic function object, the expansion of |
| 1684 | @code{define-method} will include a call to @code{define-generic}. If |
| 1685 | @var{generic} is @code{(setter @var{generic-with-setter})}, where |
| 1686 | @var{generic-with-setter} is a variable which is not yet bound to a |
| 1687 | generic-with-setter object, the expansion will include a call to |
| 1688 | @code{define-accessor}. |
| 1689 | |
| 1690 | Each @var{parameter} must be either a symbol or a two-element list |
| 1691 | @code{(@var{symbol} @var{class})}. The symbols refer to variables in |
| 1692 | the @var{body} that will be bound to the parameters supplied by the |
| 1693 | caller when calling this method. The @var{class}es, if present, |
| 1694 | specify the possible combinations of parameters to which this method |
| 1695 | can be applied. |
| 1696 | |
| 1697 | @var{body} is the body of the method definition. |
| 1698 | @end deffn |
| 1699 | |
| 1700 | @code{define-method} expressions look a little like normal Scheme |
| 1701 | procedure definitions of the form |
| 1702 | |
| 1703 | @example |
| 1704 | (define (name formals @dots{}) . body) |
| 1705 | @end example |
| 1706 | |
| 1707 | The most important difference is that each formal parameter, apart from the |
| 1708 | possible ``rest'' argument, can be qualified by a class name: |
| 1709 | @code{@var{formal}} becomes @code{(@var{formal} @var{class})}. The |
| 1710 | meaning of this qualification is that the method being defined |
| 1711 | will only be applicable in a particular generic function invocation if |
| 1712 | the corresponding argument is an instance of @code{@var{class}} (or one of |
| 1713 | its subclasses). If more than one of the formal parameters is qualified |
| 1714 | in this way, then the method will only be applicable if each of the |
| 1715 | corresponding arguments is an instance of its respective qualifying class. |
| 1716 | |
| 1717 | Note that unqualified formal parameters act as though they are qualified |
| 1718 | by the class @code{<top>}, which GOOPS uses to mean the superclass of |
| 1719 | all valid Scheme types, including both primitive types and GOOPS classes. |
| 1720 | |
| 1721 | For example, if a generic function method is defined with |
| 1722 | @var{parameter}s @code{((s1 <square>) (n <number>))}, that method is |
| 1723 | only applicable to invocations of its generic function that have two |
| 1724 | parameters where the first parameter is an instance of the |
| 1725 | @code{<square>} class and the second parameter is a number. |
| 1726 | |
| 1727 | If a generic function is invoked with a combination of parameters for which |
| 1728 | there is no applicable method, GOOPS raises an error. For more about |
| 1729 | invocation error handling, and generic function invocation in general, |
| 1730 | see @ref{Invoking Generic Functions}. |
| 1731 | |
| 1732 | @node Method Definition Internals |
| 1733 | @subsubsection Method Definition Internals |
| 1734 | |
| 1735 | @code{define-method} |
| 1736 | |
| 1737 | @itemize @bullet |
| 1738 | @item |
| 1739 | checks the form of the first parameter, and applies the following steps |
| 1740 | to the accessor's setter if it has the @code{(setter @dots{})} form |
| 1741 | |
| 1742 | @item |
| 1743 | interpolates a call to @code{define-generic} or @code{define-accessor} |
| 1744 | if a generic function is not already defined with the supplied name |
| 1745 | |
| 1746 | @item |
| 1747 | calls @code{method} with the @var{parameter}s and @var{body}, to make a |
| 1748 | new method instance |
| 1749 | |
| 1750 | @item |
| 1751 | calls @code{add-method!} to add this method to the relevant generic |
| 1752 | function. |
| 1753 | @end itemize |
| 1754 | |
| 1755 | @deffn syntax method (parameter @dots{}) . body |
| 1756 | Make a method whose specializers are defined by the classes in |
| 1757 | @var{parameter}s and whose procedure definition is constructed from the |
| 1758 | @var{parameter} symbols and @var{body} forms. |
| 1759 | |
| 1760 | The @var{parameter} and @var{body} parameters should be as for |
| 1761 | @code{define-method} (@pxref{Basic Method Definition,, define-method}). |
| 1762 | @end deffn |
| 1763 | |
| 1764 | @code{method} |
| 1765 | |
| 1766 | @itemize @bullet |
| 1767 | @item |
| 1768 | extracts formals and specializing classes from the @var{parameter}s, |
| 1769 | defaulting the class for unspecialized parameters to @code{<top>} |
| 1770 | |
| 1771 | @item |
| 1772 | creates a closure using the formals and the @var{body} forms |
| 1773 | |
| 1774 | @item |
| 1775 | calls @code{make} with metaclass @code{<method>} and the specializers |
| 1776 | and closure using the @code{#:specializers} and @code{#:procedure} |
| 1777 | keywords. |
| 1778 | @end itemize |
| 1779 | |
| 1780 | @deffn procedure make-method specializers procedure |
| 1781 | Make a method using @var{specializers} and @var{procedure}. |
| 1782 | |
| 1783 | @var{specializers} should be a list of classes that specifies the |
| 1784 | parameter combinations to which this method will be applicable. |
| 1785 | |
| 1786 | @var{procedure} should be the closure that will applied to the generic |
| 1787 | function parameters when this method is invoked. |
| 1788 | @end deffn |
| 1789 | |
| 1790 | @code{make-method} is a simple wrapper around @code{make} with metaclass |
| 1791 | @code{<method>}. |
| 1792 | |
| 1793 | @deffn generic add-method! target method |
| 1794 | Generic function for adding method @var{method} to @var{target}. |
| 1795 | @end deffn |
| 1796 | |
| 1797 | @deffn method add-method! (generic <generic>) (method <method>) |
| 1798 | Add method @var{method} to the generic function @var{generic}. |
| 1799 | @end deffn |
| 1800 | |
| 1801 | @deffn method add-method! (proc <procedure>) (method <method>) |
| 1802 | If @var{proc} is a procedure with generic capability (@pxref{Extending |
| 1803 | Guiles Primitives,, generic-capability?}), upgrade it to a |
| 1804 | primitive generic and add @var{method} to its generic function |
| 1805 | definition. |
| 1806 | @end deffn |
| 1807 | |
| 1808 | @deffn method add-method! (pg <primitive-generic>) (method <method>) |
| 1809 | Add method @var{method} to the generic function definition of @var{pg}. |
| 1810 | |
| 1811 | Implementation: @code{(add-method! (primitive-generic-generic pg) method)}. |
| 1812 | @end deffn |
| 1813 | |
| 1814 | @deffn method add-method! (whatever <top>) (method <method>) |
| 1815 | Raise an error indicating that @var{whatever} is not a valid generic |
| 1816 | function. |
| 1817 | @end deffn |
| 1818 | |
| 1819 | @node Invoking Generic Functions |
| 1820 | @subsection Invoking Generic Functions |
| 1821 | |
| 1822 | When a variable with a generic function definition appears as the first |
| 1823 | element of a list that is being evaluated, the Guile evaluator tries |
| 1824 | to apply the generic function to the arguments obtained by evaluating |
| 1825 | the remaining elements of the list. [ *fixme* How do I put this in a |
| 1826 | more Schemely and less Lispy way? ] |
| 1827 | |
| 1828 | Usually a generic function contains several method definitions, with |
| 1829 | varying degrees of formal parameter specialization (@pxref{Basic |
| 1830 | Method Definition,, define-method}). So it is necessary to sort these |
| 1831 | methods by specificity with respect to the supplied arguments, and then |
| 1832 | apply the most specific method definition. Less specific methods |
| 1833 | may be applied subsequently if a method that is being applied calls |
| 1834 | @code{next-method}. |
| 1835 | |
| 1836 | @menu |
| 1837 | * Determining Which Methods to Apply:: |
| 1838 | * Handling Invocation Errors:: |
| 1839 | @end menu |
| 1840 | |
| 1841 | @node Determining Which Methods to Apply |
| 1842 | @subsubsection Determining Which Methods to Apply |
| 1843 | |
| 1844 | [ *fixme* Sorry - this is the area of GOOPS that I understand least of |
| 1845 | all, so I'm afraid I have to pass on this section. Would some other |
| 1846 | kind person consider filling it in? ] |
| 1847 | |
| 1848 | @deffn generic apply-generic |
| 1849 | @deffnx method apply-generic (gf <generic>) args |
| 1850 | @end deffn |
| 1851 | |
| 1852 | @deffn generic compute-applicable-methods |
| 1853 | @deffnx method compute-applicable-methods (gf <generic>) args |
| 1854 | @end deffn |
| 1855 | |
| 1856 | @deffn generic sort-applicable-methods |
| 1857 | @deffnx method sort-applicable-methods (gf <generic>) methods args |
| 1858 | @end deffn |
| 1859 | |
| 1860 | @deffn generic method-more-specific? |
| 1861 | @deffnx method method-more-specific? (m1 <method>) (m2 <method>) args |
| 1862 | @end deffn |
| 1863 | |
| 1864 | @deffn generic apply-method |
| 1865 | @deffnx method apply-method (gf <generic>) methods build-next args |
| 1866 | @end deffn |
| 1867 | |
| 1868 | @deffn generic apply-methods |
| 1869 | @deffnx method apply-methods (gf <generic>) (l <list>) args |
| 1870 | @end deffn |
| 1871 | |
| 1872 | @node Handling Invocation Errors |
| 1873 | @subsubsection Handling Invocation Errors |
| 1874 | |
| 1875 | @deffn generic no-method |
| 1876 | @deffnx method no-method (gf <generic>) args |
| 1877 | When an application invokes a generic function, and no methods at all |
| 1878 | have been defined for that generic function, GOOPS calls the |
| 1879 | @code{no-method} generic function. The default method calls |
| 1880 | @code{goops-error} with an appropriate message. |
| 1881 | @end deffn |
| 1882 | |
| 1883 | @deffn generic no-applicable-method |
| 1884 | @deffnx method no-applicable-method (gf <generic>) args |
| 1885 | When an application applies a generic function to a set of arguments, |
| 1886 | and no methods have been defined for those argument types, GOOPS calls |
| 1887 | the @code{no-applicable-method} generic function. The default method |
| 1888 | calls @code{goops-error} with an appropriate message. |
| 1889 | @end deffn |
| 1890 | |
| 1891 | @deffn generic no-next-method |
| 1892 | @deffnx method no-next-method (gf <generic>) args |
| 1893 | When a generic function method calls @code{(next-method)} to invoke the |
| 1894 | next less specialized method for that generic function, and no less |
| 1895 | specialized methods have been defined for the current generic function |
| 1896 | arguments, GOOPS calls the @code{no-next-method} generic function. The |
| 1897 | default method calls @code{goops-error} with an appropriate message. |
| 1898 | @end deffn |
| 1899 | |
| 1900 | @node Redefining a Class |
| 1901 | @subsection Redefining a Class |
| 1902 | |
| 1903 | Suppose that a class @code{<my-class>} is defined using @code{define-class} |
| 1904 | (@pxref{Basic Class Definition,, define-class}), with slots that have |
| 1905 | accessor functions, and that an application has created several instances |
| 1906 | of @code{<my-class>} using @code{make} (@pxref{Basic Instance Creation,, |
| 1907 | make}). What then happens if @code{<my-class>} is redefined by calling |
| 1908 | @code{define-class} again? |
| 1909 | |
| 1910 | @menu |
| 1911 | * Default Class Redefinition Behaviour:: |
| 1912 | * Customizing Class Redefinition:: |
| 1913 | @end menu |
| 1914 | |
| 1915 | @node Default Class Redefinition Behaviour |
| 1916 | @subsubsection Default Class Redefinition Behaviour |
| 1917 | |
| 1918 | GOOPS' default answer to this question is as follows. |
| 1919 | |
| 1920 | @itemize @bullet |
| 1921 | @item |
| 1922 | All existing direct instances of @code{<my-class>} are converted to be |
| 1923 | instances of the new class. This is achieved by preserving the values |
| 1924 | of slots that exist in both the old and new definitions, and initializing the |
| 1925 | values of new slots in the usual way (@pxref{Basic Instance Creation,, |
| 1926 | make}). |
| 1927 | |
| 1928 | @item |
| 1929 | All existing subclasses of @code{<my-class>} are redefined, as though |
| 1930 | the @code{define-class} expressions that defined them were re-evaluated |
| 1931 | following the redefinition of @code{<my-class>}, and the class |
| 1932 | redefinition process described here is applied recursively to the |
| 1933 | redefined subclasses. |
| 1934 | |
| 1935 | @item |
| 1936 | Once all of its instances and subclasses have been updated, the class |
| 1937 | metaobject previously bound to the variable @code{<my-class>} is no |
| 1938 | longer needed and so can be allowed to be garbage collected. |
| 1939 | @end itemize |
| 1940 | |
| 1941 | To keep things tidy, GOOPS also needs to do a little housekeeping on |
| 1942 | methods that are associated with the redefined class. |
| 1943 | |
| 1944 | @itemize @bullet |
| 1945 | @item |
| 1946 | Slot accessor methods for slots in the old definition should be removed |
| 1947 | from their generic functions. They will be replaced by accessor methods |
| 1948 | for the slots of the new class definition. |
| 1949 | |
| 1950 | @item |
| 1951 | Any generic function method that uses the old @code{<my-class>} metaobject |
| 1952 | as one of its formal parameter specializers must be updated to refer to |
| 1953 | the new @code{<my-class>} metaobject. (Whenever a new generic function |
| 1954 | method is defined, @code{define-method} adds the method to a list stored |
| 1955 | in the class metaobject for each class used as a formal parameter |
| 1956 | specializer, so it is easy to identify all the methods that must be |
| 1957 | updated when a class is redefined.) |
| 1958 | @end itemize |
| 1959 | |
| 1960 | If this class redefinition strategy strikes you as rather counter-intuitive, |
| 1961 | bear in mind that it is derived from similar behaviour in other object |
| 1962 | systems such as CLOS, and that experience in those systems has shown it to be |
| 1963 | very useful in practice. |
| 1964 | |
| 1965 | Also bear in mind that, like most of GOOPS' default behaviour, it can |
| 1966 | be customized@dots{} |
| 1967 | |
| 1968 | @node Customizing Class Redefinition |
| 1969 | @subsubsection Customizing Class Redefinition |
| 1970 | |
| 1971 | When @code{define-class} notices that a class is being redefined, |
| 1972 | it constructs the new class metaobject as usual, and then invokes the |
| 1973 | @code{class-redefinition} generic function with the old and new classes |
| 1974 | as arguments. Therefore, if the old or new classes have metaclasses |
| 1975 | other than the default @code{<class>}, class redefinition behaviour can |
| 1976 | be customized by defining a @code{class-redefinition} method that is |
| 1977 | specialized for the relevant metaclasses. |
| 1978 | |
| 1979 | @deffn generic class-redefinition |
| 1980 | Handle the class redefinition from @var{old-class} to @var{new-class}, |
| 1981 | and return the new class metaobject that should be bound to the |
| 1982 | variable specified by @code{define-class}'s first argument. |
| 1983 | @end deffn |
| 1984 | |
| 1985 | @deffn method class-redefinition (old-class <class>) (new-class <class>) |
| 1986 | Implements GOOPS' default class redefinition behaviour, as described in |
| 1987 | @ref{Default Class Redefinition Behaviour}. Returns the metaobject |
| 1988 | for the new class definition. |
| 1989 | @end deffn |
| 1990 | |
| 1991 | An alternative class redefinition strategy could be to leave all |
| 1992 | existing instances as instances of the old class, but accepting that the |
| 1993 | old class is now ``nameless'', since its name has been taken over by the |
| 1994 | new definition. In this strategy, any existing subclasses could also |
| 1995 | be left as they are, on the understanding that they inherit from a nameless |
| 1996 | superclass. |
| 1997 | |
| 1998 | This strategy is easily implemented in GOOPS, by defining a new metaclass, |
| 1999 | that will be used as the metaclass for all classes to which the strategy |
| 2000 | should apply, and then defining a @code{class-redefinition} method that |
| 2001 | is specialized for this metaclass: |
| 2002 | |
| 2003 | @example |
| 2004 | (define-class <can-be-nameless> (<class>)) |
| 2005 | |
| 2006 | (define-method (class-redefinition (old <can-be-nameless>) (new <class>)) |
| 2007 | new) |
| 2008 | @end example |
| 2009 | |
| 2010 | When customization can be as easy as this, aren't you glad that GOOPS |
| 2011 | implements the far more difficult strategy as its default! |
| 2012 | |
| 2013 | Finally, note that, if @code{class-redefinition} itself is not customized, |
| 2014 | the default @code{class-redefinition} method invokes three further |
| 2015 | generic functions that could be individually customized: |
| 2016 | |
| 2017 | @itemize @bullet |
| 2018 | @item |
| 2019 | (remove-class-accessors! @var{old-class}) |
| 2020 | |
| 2021 | @item |
| 2022 | (update-direct-method! @var{method} @var{old-class} @var{new-class}) |
| 2023 | |
| 2024 | @item |
| 2025 | (update-direct-subclass! @var{subclass} @var{old-class} @var{new-class}) |
| 2026 | @end itemize |
| 2027 | |
| 2028 | and the default methods for these generic functions invoke further |
| 2029 | generic functions, and so on@dots{} The detailed protocol for all of these |
| 2030 | is described in @ref{MOP Specification}. |
| 2031 | |
| 2032 | @node Changing the Class of an Instance |
| 2033 | @subsection Changing the Class of an Instance |
| 2034 | |
| 2035 | You can change the class of an existing instance by invoking the |
| 2036 | generic function @code{change-class} with two arguments: the instance |
| 2037 | and the new class. |
| 2038 | |
| 2039 | @deffn generic change-class |
| 2040 | @end deffn |
| 2041 | |
| 2042 | The default method for @code{change-class} decides how to implement the |
| 2043 | change of class by looking at the slot definitions for the instance's |
| 2044 | existing class and for the new class. If the new class has slots with |
| 2045 | the same name as slots in the existing class, the values for those slots |
| 2046 | are preserved. Slots that are present only in the existing class are |
| 2047 | discarded. Slots that are present only in the new class are initialized |
| 2048 | using the corresponding slot definition's init function (@pxref{Classes,, |
| 2049 | slot-init-function}). |
| 2050 | |
| 2051 | @deffn {method} change-class (obj <object>) (new <class>) |
| 2052 | Modify instance @var{obj} to make it an instance of class @var{new}. |
| 2053 | |
| 2054 | The value of each of @var{obj}'s slots is preserved only if a similarly named |
| 2055 | slot exists in @var{new}; any other slot values are discarded. |
| 2056 | |
| 2057 | The slots in @var{new} that do not correspond to any of @var{obj}'s |
| 2058 | pre-existing slots are initialized according to @var{new}'s slot definitions' |
| 2059 | init functions. |
| 2060 | @end deffn |
| 2061 | |
| 2062 | Customized change of class behaviour can be implemented by defining |
| 2063 | @code{change-class} methods that are specialized either by the class |
| 2064 | of the instances to be modified or by the metaclass of the new class. |
| 2065 | |
| 2066 | When a class is redefined (@pxref{Redefining a Class}), and the default |
| 2067 | class redefinition behaviour is not overridden, GOOPS (eventually) |
| 2068 | invokes the @code{change-class} generic function for each existing |
| 2069 | instance of the redefined class. |
| 2070 | |
| 2071 | @node Introspection |
| 2072 | @subsection Introspection |
| 2073 | |
| 2074 | @dfn{Introspection}, also known as @dfn{reflection}, is the name given |
| 2075 | to the ability to obtain information dynamically about GOOPS metaobjects. |
| 2076 | It is perhaps best illustrated by considering an object oriented language |
| 2077 | that does not provide any introspection, namely C++. |
| 2078 | |
| 2079 | Nothing in C++ allows a running program to obtain answers to the following |
| 2080 | types of question: |
| 2081 | |
| 2082 | @itemize @bullet |
| 2083 | @item |
| 2084 | What are the data members of this object or class? |
| 2085 | |
| 2086 | @item |
| 2087 | What classes does this class inherit from? |
| 2088 | |
| 2089 | @item |
| 2090 | Is this method call virtual or non-virtual? |
| 2091 | |
| 2092 | @item |
| 2093 | If I invoke @code{Employee::adjustHoliday()}, what class contains the |
| 2094 | @code{adjustHoliday()} method that will be applied? |
| 2095 | @end itemize |
| 2096 | |
| 2097 | In C++, answers to such questions can only be determined by looking at |
| 2098 | the source code, if you have access to it. GOOPS, on the other hand, |
| 2099 | includes procedures that allow answers to these questions --- or their |
| 2100 | GOOPS equivalents --- to be obtained dynamically, at run time. |
| 2101 | |
| 2102 | @menu |
| 2103 | * Classes:: |
| 2104 | * Slots:: |
| 2105 | * Instances:: |
| 2106 | * Generic Functions:: |
| 2107 | * Generic Function Methods:: |
| 2108 | @end menu |
| 2109 | |
| 2110 | @node Classes |
| 2111 | @subsubsection Classes |
| 2112 | |
| 2113 | @deffn {primitive procedure} class-name class |
| 2114 | Return the name of class @var{class}. |
| 2115 | This is the value of the @var{class} metaobject's @code{name} slot. |
| 2116 | @end deffn |
| 2117 | |
| 2118 | @deffn {primitive procedure} class-direct-supers class |
| 2119 | Return a list containing the direct superclasses of @var{class}. |
| 2120 | This is the value of the @var{class} metaobject's |
| 2121 | @code{direct-supers} slot. |
| 2122 | @end deffn |
| 2123 | |
| 2124 | @deffn {primitive procedure} class-direct-slots class |
| 2125 | Return a list containing the slot definitions of the direct slots of |
| 2126 | @var{class}. |
| 2127 | This is the value of the @var{class} metaobject's @code{direct-slots} |
| 2128 | slot. |
| 2129 | @end deffn |
| 2130 | |
| 2131 | @deffn {primitive procedure} class-direct-subclasses class |
| 2132 | Return a list containing the direct subclasses of @var{class}. |
| 2133 | This is the value of the @var{class} metaobject's |
| 2134 | @code{direct-subclasses} slot. |
| 2135 | @end deffn |
| 2136 | |
| 2137 | @deffn {primitive procedure} class-direct-methods class |
| 2138 | Return a list of all the generic function methods that use @var{class} |
| 2139 | as a formal parameter specializer. |
| 2140 | This is the value of the @var{class} metaobject's @code{direct-methods} |
| 2141 | slot. |
| 2142 | @end deffn |
| 2143 | |
| 2144 | @deffn {primitive procedure} class-precedence-list class |
| 2145 | Return the class precedence list for class @var{class} (@pxref{Class |
| 2146 | precedence list}). |
| 2147 | This is the value of the @var{class} metaobject's @code{cpl} slot. |
| 2148 | @end deffn |
| 2149 | |
| 2150 | @deffn {primitive procedure} class-slots class |
| 2151 | Return a list containing the slot definitions for all @var{class}'s slots, |
| 2152 | including any slots that are inherited from superclasses. |
| 2153 | This is the value of the @var{class} metaobject's @code{slots} slot. |
| 2154 | @end deffn |
| 2155 | |
| 2156 | @deffn {primitive procedure} class-environment class |
| 2157 | Return the value of @var{class}'s @code{environment} slot. |
| 2158 | [ *fixme* I don't know what this value is used for. ] |
| 2159 | @end deffn |
| 2160 | |
| 2161 | @deffn procedure class-subclasses class |
| 2162 | Return a list of all subclasses of @var{class}. |
| 2163 | @end deffn |
| 2164 | |
| 2165 | @deffn procedure class-methods class |
| 2166 | Return a list of all methods that use @var{class} or a subclass of |
| 2167 | @var{class} as one of its formal parameter specializers. |
| 2168 | @end deffn |
| 2169 | |
| 2170 | @node Slots |
| 2171 | @subsubsection Slots |
| 2172 | |
| 2173 | @deffn procedure class-slot-definition class slot-name |
| 2174 | Return the slot definition for the slot named @var{slot-name} in class |
| 2175 | @var{class}. @var{slot-name} should be a symbol. |
| 2176 | @end deffn |
| 2177 | |
| 2178 | @deffn procedure slot-definition-name slot-def |
| 2179 | Extract and return the slot name from @var{slot-def}. |
| 2180 | @end deffn |
| 2181 | |
| 2182 | @deffn procedure slot-definition-options slot-def |
| 2183 | Extract and return the slot options from @var{slot-def}. |
| 2184 | @end deffn |
| 2185 | |
| 2186 | @deffn procedure slot-definition-allocation slot-def |
| 2187 | Extract and return the slot allocation option from @var{slot-def}. This |
| 2188 | is the value of the @code{#:allocation} keyword (@pxref{Slot Options,, |
| 2189 | allocation}), or @code{#:instance} if the @code{#:allocation} keyword is |
| 2190 | absent. |
| 2191 | @end deffn |
| 2192 | |
| 2193 | @deffn procedure slot-definition-getter slot-def |
| 2194 | Extract and return the slot getter option from @var{slot-def}. This is |
| 2195 | the value of the @code{#:getter} keyword (@pxref{Slot Options,, |
| 2196 | getter}), or @code{#f} if the @code{#:getter} keyword is absent. |
| 2197 | @end deffn |
| 2198 | |
| 2199 | @deffn procedure slot-definition-setter slot-def |
| 2200 | Extract and return the slot setter option from @var{slot-def}. This is |
| 2201 | the value of the @code{#:setter} keyword (@pxref{Slot Options,, |
| 2202 | setter}), or @code{#f} if the @code{#:setter} keyword is absent. |
| 2203 | @end deffn |
| 2204 | |
| 2205 | @deffn procedure slot-definition-accessor slot-def |
| 2206 | Extract and return the slot accessor option from @var{slot-def}. This |
| 2207 | is the value of the @code{#:accessor} keyword (@pxref{Slot Options,, |
| 2208 | accessor}), or @code{#f} if the @code{#:accessor} keyword is absent. |
| 2209 | @end deffn |
| 2210 | |
| 2211 | @deffn procedure slot-definition-init-value slot-def |
| 2212 | Extract and return the slot init-value option from @var{slot-def}. This |
| 2213 | is the value of the @code{#:init-value} keyword (@pxref{Slot Options,, |
| 2214 | init-value}), or the unbound value if the @code{#:init-value} keyword is |
| 2215 | absent. |
| 2216 | @end deffn |
| 2217 | |
| 2218 | @deffn procedure slot-definition-init-form slot-def |
| 2219 | Extract and return the slot init-form option from @var{slot-def}. This |
| 2220 | is the value of the @code{#:init-form} keyword (@pxref{Slot Options,, |
| 2221 | init-form}), or the unbound value if the @code{#:init-form} keyword is |
| 2222 | absent. |
| 2223 | @end deffn |
| 2224 | |
| 2225 | @deffn procedure slot-definition-init-thunk slot-def |
| 2226 | Extract and return the slot init-thunk option from @var{slot-def}. This |
| 2227 | is the value of the @code{#:init-thunk} keyword (@pxref{Slot Options,, |
| 2228 | init-thunk}), or @code{#f} if the @code{#:init-thunk} keyword is absent. |
| 2229 | @end deffn |
| 2230 | |
| 2231 | @deffn procedure slot-definition-init-keyword slot-def |
| 2232 | Extract and return the slot init-keyword option from @var{slot-def}. |
| 2233 | This is the value of the @code{#:init-keyword} keyword (@pxref{Slot |
| 2234 | Options,, init-keyword}), or @code{#f} if the @code{#:init-keyword} |
| 2235 | keyword is absent. |
| 2236 | @end deffn |
| 2237 | |
| 2238 | @deffn procedure slot-init-function class slot-name |
| 2239 | Return the initialization function for the slot named @var{slot-name} in |
| 2240 | class @var{class}. @var{slot-name} should be a symbol. |
| 2241 | |
| 2242 | The returned initialization function incorporates the effects of the |
| 2243 | standard @code{#:init-thunk}, @code{#:init-form} and @code{#:init-value} |
| 2244 | slot options. These initializations can be overridden by the |
| 2245 | @code{#:init-keyword} slot option or by a specialized @code{initialize} |
| 2246 | method, so, in general, the function returned by |
| 2247 | @code{slot-init-function} may be irrelevant. For a fuller discussion, |
| 2248 | see @ref{Slot Options,, init-value}. |
| 2249 | @end deffn |
| 2250 | |
| 2251 | @node Instances |
| 2252 | @subsubsection Instances |
| 2253 | |
| 2254 | @deffn {primitive procedure} class-of value |
| 2255 | Return the GOOPS class of any Scheme @var{value}. |
| 2256 | @end deffn |
| 2257 | |
| 2258 | @deffn {primitive procedure} instance? object |
| 2259 | Return @code{#t} if @var{object} is any GOOPS instance, otherwise |
| 2260 | @code{#f}. |
| 2261 | @end deffn |
| 2262 | |
| 2263 | @deffn procedure is-a? object class |
| 2264 | Return @code{#t} if @var{object} is an instance of @var{class} or one of |
| 2265 | its subclasses. |
| 2266 | @end deffn |
| 2267 | |
| 2268 | Implementation notes: @code{is-a?} uses @code{class-of} and |
| 2269 | @code{class-precedence-list} to obtain the class precedence list for |
| 2270 | @var{object}. |
| 2271 | |
| 2272 | @node Generic Functions |
| 2273 | @subsubsection Generic Functions |
| 2274 | |
| 2275 | @deffn {primitive procedure} generic-function-name gf |
| 2276 | Return the name of generic function @var{gf}. |
| 2277 | @end deffn |
| 2278 | |
| 2279 | @deffn {primitive procedure} generic-function-methods gf |
| 2280 | Return a list of the methods of generic function @var{gf}. |
| 2281 | This is the value of the @var{gf} metaobject's @code{methods} slot. |
| 2282 | @end deffn |
| 2283 | |
| 2284 | @node Generic Function Methods |
| 2285 | @subsubsection Generic Function Methods |
| 2286 | |
| 2287 | @deffn {primitive procedure} method-generic-function method |
| 2288 | Return the generic function that @var{method} belongs to. |
| 2289 | This is the value of the @var{method} metaobject's |
| 2290 | @code{generic-function} slot. |
| 2291 | @end deffn |
| 2292 | |
| 2293 | @deffn {primitive procedure} method-specializers method |
| 2294 | Return a list of @var{method}'s formal parameter specializers . |
| 2295 | This is the value of the @var{method} metaobject's |
| 2296 | @code{specializers} slot. |
| 2297 | @end deffn |
| 2298 | |
| 2299 | @deffn {primitive procedure} method-procedure method |
| 2300 | Return the procedure that implements @var{method}. |
| 2301 | This is the value of the @var{method} metaobject's |
| 2302 | @code{procedure} slot. |
| 2303 | @end deffn |
| 2304 | |
| 2305 | @deffn generic method-source |
| 2306 | @deffnx method method-source (m <method>) |
| 2307 | Return an expression that prints to show the definition of method |
| 2308 | @var{m}. |
| 2309 | |
| 2310 | @example |
| 2311 | (define-generic cube) |
| 2312 | |
| 2313 | (define-method (cube (n <number>)) |
| 2314 | (* n n n)) |
| 2315 | |
| 2316 | (map method-source (generic-function-methods cube)) |
| 2317 | @result{} |
| 2318 | ((method ((n <number>)) (* n n n))) |
| 2319 | @end example |
| 2320 | @end deffn |
| 2321 | |
| 2322 | @node Miscellaneous Functions |
| 2323 | @subsection Miscellaneous Functions |
| 2324 | |
| 2325 | @menu |
| 2326 | * Administrative Functions:: |
| 2327 | * GOOPS Error Handling:: |
| 2328 | * Object Comparisons:: |
| 2329 | * Cloning Objects:: |
| 2330 | * Write and Display:: |
| 2331 | @end menu |
| 2332 | |
| 2333 | @node Administrative Functions |
| 2334 | @subsubsection Administration Functions |
| 2335 | |
| 2336 | This section describes administrative, non-technical GOOPS functions. |
| 2337 | |
| 2338 | @deffn primitive goops-version |
| 2339 | Return the current GOOPS version as a string, for example ``0.2''. |
| 2340 | @end deffn |
| 2341 | |
| 2342 | @node GOOPS Error Handling |
| 2343 | @subsubsection Error Handling |
| 2344 | |
| 2345 | The procedure @code{goops-error} is called to raise an appropriate error |
| 2346 | by the default methods of the following generic functions: |
| 2347 | |
| 2348 | @itemize @bullet |
| 2349 | @item |
| 2350 | @code{slot-missing} (@pxref{Handling Slot Access Errors,, slot-missing}) |
| 2351 | |
| 2352 | @item |
| 2353 | @code{slot-unbound} (@pxref{Handling Slot Access Errors,, slot-unbound}) |
| 2354 | |
| 2355 | @item |
| 2356 | @code{no-method} (@pxref{Handling Invocation Errors,, no-method}) |
| 2357 | |
| 2358 | @item |
| 2359 | @code{no-applicable-method} (@pxref{Handling Invocation Errors,, |
| 2360 | no-applicable-method}) |
| 2361 | |
| 2362 | @item |
| 2363 | @code{no-next-method} (@pxref{Handling Invocation Errors,, |
| 2364 | no-next-method}) |
| 2365 | @end itemize |
| 2366 | |
| 2367 | If you customize these functions for particular classes or metaclasses, |
| 2368 | you may still want to use @code{goops-error} to signal any error |
| 2369 | conditions that you detect. |
| 2370 | |
| 2371 | @deffn procedure goops-error format-string . args |
| 2372 | Raise an error with key @code{goops-error} and error message constructed |
| 2373 | from @var{format-string} and @var{args}. Error message formatting is |
| 2374 | as done by @code{scm-error}. |
| 2375 | @end deffn |
| 2376 | |
| 2377 | @node Object Comparisons |
| 2378 | @subsubsection Object Comparisons |
| 2379 | |
| 2380 | @deffn generic eqv? |
| 2381 | @deffnx method eqv? ((x <top>) (y <top>)) |
| 2382 | @deffnx generic equal? |
| 2383 | @deffnx method equal? ((x <top>) (y <top>)) |
| 2384 | @deffnx generic = |
| 2385 | @deffnx method = ((x <number>) (y <number>)) |
| 2386 | Generic functions and default (unspecialized) methods for comparing two |
| 2387 | GOOPS objects. |
| 2388 | |
| 2389 | The default method for @code{eqv?} returns @code{#t} for all values |
| 2390 | that are equal in the sense defined by R5RS and the Guile reference |
| 2391 | manual, otherwise @code{#f}. The default method for @code{equal?} |
| 2392 | returns @code{#t} or @code{#f} in the sense defined by R5RS and the |
| 2393 | Guile reference manual. If no such comparison is defined, |
| 2394 | @code{equal?} returns the result of a call to @code{eqv?}. The |
| 2395 | default method for = returns @code{#t} if @var{x} and @var{y} are |
| 2396 | numerically equal, otherwise @code{#f}. |
| 2397 | |
| 2398 | Application class authors may wish to define specialized methods for |
| 2399 | @code{eqv?}, @code{equal?} and @code{=} that compare instances of the |
| 2400 | same class for equality in whatever sense is useful to the |
| 2401 | application. Such methods will only be called if the arguments have |
| 2402 | the same class and the result of the comparison isn't defined by R5RS |
| 2403 | and the Guile reference manual. |
| 2404 | @end deffn |
| 2405 | |
| 2406 | @node Cloning Objects |
| 2407 | @subsubsection Cloning Objects |
| 2408 | |
| 2409 | @deffn generic shallow-clone |
| 2410 | @deffnx method shallow-clone (self <object>) |
| 2411 | Return a ``shallow'' clone of @var{self}. The default method makes a |
| 2412 | shallow clone by allocating a new instance and copying slot values from |
| 2413 | self to the new instance. Each slot value is copied either as an |
| 2414 | immediate value or by reference. |
| 2415 | @end deffn |
| 2416 | |
| 2417 | @deffn generic deep-clone |
| 2418 | @deffnx method deep-clone (self <object>) |
| 2419 | Return a ``deep'' clone of @var{self}. The default method makes a deep |
| 2420 | clone by allocating a new instance and copying or cloning slot values |
| 2421 | from self to the new instance. If a slot value is an instance |
| 2422 | (satisfies @code{instance?}), it is cloned by calling @code{deep-clone} |
| 2423 | on that value. Other slot values are copied either as immediate values |
| 2424 | or by reference. |
| 2425 | @end deffn |
| 2426 | |
| 2427 | @node Write and Display |
| 2428 | @subsubsection Write and Display |
| 2429 | |
| 2430 | @deffn {primitive generic} write object port |
| 2431 | @deffnx {primitive generic} display object port |
| 2432 | When GOOPS is loaded, @code{write} and @code{display} become generic |
| 2433 | functions with special methods for printing |
| 2434 | |
| 2435 | @itemize @bullet |
| 2436 | @item |
| 2437 | objects - instances of the class @code{<object>} |
| 2438 | |
| 2439 | @item |
| 2440 | foreign objects - instances of the class @code{<foreign-object>} |
| 2441 | |
| 2442 | @item |
| 2443 | classes - instances of the class @code{<class>} |
| 2444 | |
| 2445 | @item |
| 2446 | generic functions - instances of the class @code{<generic>} |
| 2447 | |
| 2448 | @item |
| 2449 | methods - instances of the class @code{<method>}. |
| 2450 | @end itemize |
| 2451 | |
| 2452 | @code{write} and @code{display} print non-GOOPS values in the same way |
| 2453 | as the Guile primitive @code{write} and @code{display} functions. |
| 2454 | @end deffn |
| 2455 | |
| 2456 | @node MOP Specification |
| 2457 | @section MOP Specification |
| 2458 | |
| 2459 | For an introduction to metaobjects and the metaobject protocol, |
| 2460 | see @ref{Metaobjects and the Metaobject Protocol}. |
| 2461 | |
| 2462 | The aim of the MOP specification in this chapter is to specify all the |
| 2463 | customizable generic function invocations that can be made by the standard |
| 2464 | GOOPS syntax, procedures and methods, and to explain the protocol for |
| 2465 | customizing such invocations. |
| 2466 | |
| 2467 | A generic function invocation is customizable if the types of the arguments |
| 2468 | to which it is applied are not all determined by the lexical context in |
| 2469 | which the invocation appears. For example, |
| 2470 | |
| 2471 | @itemize @bullet |
| 2472 | @item |
| 2473 | the @code{(initialize @var{instance} @var{initargs})} invocation in the |
| 2474 | default @code{make-instance} method is customizable, because the type of the |
| 2475 | @code{@var{instance}} argument is determined by the class that was passed to |
| 2476 | @code{make-instance}. |
| 2477 | |
| 2478 | @item |
| 2479 | the @code{(make <generic> #:name ',name)} invocation in @code{define-generic} |
| 2480 | is not customizable, because all of its arguments have lexically determined |
| 2481 | types. |
| 2482 | @end itemize |
| 2483 | |
| 2484 | When using this rule to decide whether a given generic function invocation |
| 2485 | is customizable, we ignore arguments that are expected to be handled in |
| 2486 | method definitions as a single ``rest'' list argument. |
| 2487 | |
| 2488 | For each customizable generic function invocation, the @dfn{invocation |
| 2489 | protocol} is explained by specifying |
| 2490 | |
| 2491 | @itemize @bullet |
| 2492 | @item |
| 2493 | what, conceptually, the applied method is intended to do |
| 2494 | |
| 2495 | @item |
| 2496 | what assumptions, if any, the caller makes about the applied method's side |
| 2497 | effects |
| 2498 | |
| 2499 | @item |
| 2500 | what the caller expects to get as the applied method's return value. |
| 2501 | @end itemize |
| 2502 | |
| 2503 | @menu |
| 2504 | * Class Definition:: |
| 2505 | * Instance Creation:: |
| 2506 | * Class Redefinition:: |
| 2507 | * Method Definition:: |
| 2508 | * Generic Function Invocation:: |
| 2509 | @end menu |
| 2510 | |
| 2511 | @node Class Definition |
| 2512 | @subsection Class Definition |
| 2513 | |
| 2514 | @code{define-class} (syntax) |
| 2515 | |
| 2516 | @itemize @bullet |
| 2517 | @item |
| 2518 | @code{class} (syntax) |
| 2519 | |
| 2520 | @itemize @bullet |
| 2521 | @item |
| 2522 | @code{make-class} (procedure) |
| 2523 | |
| 2524 | @itemize @bullet |
| 2525 | @item |
| 2526 | @code{make @var{metaclass} @dots{}} (generic) |
| 2527 | |
| 2528 | @var{metaclass} is the metaclass of the class being defined, either |
| 2529 | taken from the @code{#:metaclass} class option or computed by |
| 2530 | @code{ensure-metaclass}. The applied method must create and return the |
| 2531 | fully initialized class metaobject for the new class definition. |
| 2532 | @end itemize |
| 2533 | |
| 2534 | @end itemize |
| 2535 | |
| 2536 | @item |
| 2537 | @code{class-redefinition @var{old-class} @var{new-class}} (generic) |
| 2538 | |
| 2539 | @code{define-class} calls @code{class-redefinition} if the variable |
| 2540 | specified by its first argument already held a GOOPS class definition. |
| 2541 | @var{old-class} and @var{new-class} are the old and new class metaobjects. |
| 2542 | The applied method should perform whatever is necessary to handle the |
| 2543 | redefinition, and should return the class metaobject that is to be bound |
| 2544 | to @code{define-class}'s variable. The default class redefinition |
| 2545 | protocol is described in @ref{Class Redefinition}. |
| 2546 | @end itemize |
| 2547 | |
| 2548 | The @code{(make @var{metaclass} @dots{})} invocation above will create |
| 2549 | an class metaobject with metaclass @var{metaclass}. By default, this |
| 2550 | metaobject will be initialized by the @code{initialize} method that is |
| 2551 | specialized for instances of type @code{<class>}. |
| 2552 | |
| 2553 | @code{initialize <class> @var{initargs}} (method) |
| 2554 | |
| 2555 | @itemize @bullet |
| 2556 | @item |
| 2557 | @code{compute-cpl @var{class}} (generic) |
| 2558 | |
| 2559 | The applied method should compute and return the class precedence list |
| 2560 | for @var{class} as a list of class metaobjects. When @code{compute-cpl} |
| 2561 | is called, the following @var{class} metaobject slots have all been |
| 2562 | initialized: @code{name}, @code{direct-supers}, @code{direct-slots}, |
| 2563 | @code{direct-subclasses} (empty), @code{direct-methods}. The value |
| 2564 | returned by @code{compute-cpl} will be stored in the @code{cpl} slot. |
| 2565 | |
| 2566 | @item |
| 2567 | @code{compute-slots @var{class}} (generic) |
| 2568 | |
| 2569 | The applied method should compute and return the slots (union of direct |
| 2570 | and inherited) for @var{class} as a list of slot definitions. When |
| 2571 | @code{compute-slots} is called, all the @var{class} metaobject slots |
| 2572 | mentioned for @code{compute-cpl} have been initialized, plus the |
| 2573 | following: @code{cpl}, @code{redefined} (@code{#f}), @code{environment}. |
| 2574 | The value returned by @code{compute-slots} will be stored in the |
| 2575 | @code{slots} slot. |
| 2576 | |
| 2577 | @item |
| 2578 | @code{compute-get-n-set @var{class} @var{slot-def}} (generic) |
| 2579 | |
| 2580 | @code{initialize} calls @code{compute-get-n-set} for each slot computed |
| 2581 | by @code{compute-slots}. The applied method should compute and return a |
| 2582 | pair of closures that, respectively, get and set the value of the specified |
| 2583 | slot. The get closure should have arity 1 and expect a single argument |
| 2584 | that is the instance whose slot value is to be retrieved. The set closure |
| 2585 | should have arity 2 and expect two arguments, where the first argument is |
| 2586 | the instance whose slot value is to be set and the second argument is the |
| 2587 | new value for that slot. The closures should be returned in a two element |
| 2588 | list: @code{(list @var{get} @var{set})}. |
| 2589 | |
| 2590 | The closures returned by @code{compute-get-n-set} are stored as part of |
| 2591 | the value of the @var{class} metaobject's @code{getters-n-setters} slot. |
| 2592 | Specifically, the value of this slot is a list with the same number of |
| 2593 | elements as there are slots in the class, and each element looks either like |
| 2594 | |
| 2595 | @example |
| 2596 | @code{(@var{slot-name-symbol} @var{init-function} . @var{index})} |
| 2597 | @end example |
| 2598 | |
| 2599 | or like |
| 2600 | |
| 2601 | @example |
| 2602 | @code{(@var{slot-name-symbol} @var{init-function} @var{get} @var{set})} |
| 2603 | @end example |
| 2604 | |
| 2605 | Where the get and set closures are replaced by @var{index}, the slot is |
| 2606 | an instance slot and @var{index} is the slot's index in the underlying |
| 2607 | structure: GOOPS knows how to get and set the value of such slots and so |
| 2608 | does not need specially constructed get and set closures. Otherwise, |
| 2609 | @var{get} and @var{set} are the closures returned by @code{compute-get-n-set}. |
| 2610 | |
| 2611 | The structure of the @code{getters-n-setters} slot value is important when |
| 2612 | understanding the next customizable generic functions that @code{initialize} |
| 2613 | calls@dots{} |
| 2614 | |
| 2615 | @item |
| 2616 | @code{compute-getter-method @var{class} @var{gns}} (generic) |
| 2617 | |
| 2618 | @code{initialize} calls @code{compute-getter-method} for each of the class's |
| 2619 | slots (as determined by @code{compute-slots}) that includes a |
| 2620 | @code{#:getter} or @code{#:accessor} slot option. @var{gns} is the |
| 2621 | element of the @var{class} metaobject's @code{getters-n-setters} slot that |
| 2622 | specifies how the slot in question is referenced and set, as described |
| 2623 | above under @code{compute-get-n-set}. The applied method should create |
| 2624 | and return a method that is specialized for instances of type @var{class} |
| 2625 | and uses the get closure to retrieve the slot's value. [ *fixme Need |
| 2626 | to insert something here about checking that the value is not unbound. ] |
| 2627 | @code{initialize} uses @code{add-method!} to add the returned method to |
| 2628 | the generic function named by the slot definition's @code{#:getter} or |
| 2629 | @code{#:accessor} option. |
| 2630 | |
| 2631 | @item |
| 2632 | @code{compute-setter-method @var{class} @var{gns}} (generic) |
| 2633 | |
| 2634 | @code{compute-setter-method} is invoked with the same arguments as |
| 2635 | @code{compute-getter-method}, for each of the class's slots that includes |
| 2636 | a @code{#:setter} or @code{#:accessor} slot option. The applied method |
| 2637 | should create and return a method that is specialized for instances of |
| 2638 | type @var{class} and uses the set closure to set the slot's value. |
| 2639 | @code{initialize} then uses @code{add-method!} to add the returned method |
| 2640 | to the generic function named by the slot definition's @code{#:setter} |
| 2641 | or @code{#:accessor} option. |
| 2642 | @end itemize |
| 2643 | |
| 2644 | @node Instance Creation |
| 2645 | @subsection Instance Creation |
| 2646 | |
| 2647 | @code{make <class> . @var{initargs}} (method) |
| 2648 | |
| 2649 | @itemize @bullet |
| 2650 | @item |
| 2651 | @code{allocate-instance @var{class} @var{initargs}} (generic) |
| 2652 | |
| 2653 | The applied @code{allocate-instance} method should allocate storage for |
| 2654 | a new instance of class @var{class} and return the uninitialized instance. |
| 2655 | |
| 2656 | @item |
| 2657 | @code{initialize @var{instance} @var{initargs}} (generic) |
| 2658 | |
| 2659 | @var{instance} is the uninitialized instance returned by |
| 2660 | @code{allocate-instance}. The applied method should initialize the new |
| 2661 | instance in whatever sense is appropriate for its class. The method's |
| 2662 | return value is ignored. |
| 2663 | @end itemize |
| 2664 | |
| 2665 | @node Class Redefinition |
| 2666 | @subsection Class Redefinition |
| 2667 | |
| 2668 | The default @code{class-redefinition} method, specialized for classes |
| 2669 | with the default metaclass @code{<class>}, has the following internal |
| 2670 | protocol. |
| 2671 | |
| 2672 | @code{class-redefinition (@var{old <class>}) (@var{new <class>})} |
| 2673 | (method) |
| 2674 | |
| 2675 | @itemize @bullet |
| 2676 | @item |
| 2677 | @code{remove-class-accessors! @var{old}} (generic) |
| 2678 | |
| 2679 | @item |
| 2680 | @code{update-direct-method! @var{method} @var{old} @var{new}} (generic) |
| 2681 | |
| 2682 | @item |
| 2683 | @code{update-direct-subclass! @var{subclass} @var{old} @var{new}} (generic) |
| 2684 | @end itemize |
| 2685 | |
| 2686 | This protocol cleans up things that the definition of the old class |
| 2687 | once changed and modifies things to work with the new class. |
| 2688 | |
| 2689 | The default @code{remove-class-accessors!} method removes the |
| 2690 | accessor methods of the old class from all classes which they |
| 2691 | specialize. |
| 2692 | |
| 2693 | The default @code{update-direct-method!} method substitutes the new |
| 2694 | class for the old in all methods specialized to the old class. |
| 2695 | |
| 2696 | The default @code{update-direct-subclass!} method invokes |
| 2697 | @code{class-redefinition} recursively to handle the redefinition of |
| 2698 | subclasses. |
| 2699 | |
| 2700 | When a class is redefined, any existing instance of the redefined class |
| 2701 | will be modified for the new class definition before the next time that |
| 2702 | any of the instance's slot is referenced or set. GOOPS modifies each |
| 2703 | instance by calling the generic function @code{change-class}. |
| 2704 | |
| 2705 | The default @code{change-class} method copies slot values from the old |
| 2706 | to the modified instance, and initializes new slots, as described in |
| 2707 | @ref{Changing the Class of an Instance}. After doing so, it makes a |
| 2708 | generic function invocation that can be used to customize the instance |
| 2709 | update algorithm. |
| 2710 | |
| 2711 | @code{change-class (@var{old-instance <object>}) (@var{new <class>})} (method) |
| 2712 | |
| 2713 | @itemize @bullet |
| 2714 | @item |
| 2715 | @code{update-instance-for-different-class @var{old-instance} @var{new-instance}} (generic) |
| 2716 | |
| 2717 | @code{change-class} invokes @code{update-instance-for-different-class} |
| 2718 | as the last thing that it does before returning. The applied method can |
| 2719 | make any further adjustments to @var{new-instance} that are required to |
| 2720 | complete or modify the change of class. The return value from the |
| 2721 | applied method is ignored. |
| 2722 | |
| 2723 | The default @code{update-instance-for-different-class} method does |
| 2724 | nothing. |
| 2725 | @end itemize |
| 2726 | |
| 2727 | @node Method Definition |
| 2728 | @subsection Method Definition |
| 2729 | |
| 2730 | @code{define-method} (syntax) |
| 2731 | |
| 2732 | @itemize @bullet |
| 2733 | @item |
| 2734 | @code{add-method! @var{target} @var{method}} (generic) |
| 2735 | |
| 2736 | @code{define-method} invokes the @code{add-method!} generic function to |
| 2737 | handle adding the new method to a variety of possible targets. GOOPS |
| 2738 | includes methods to handle @var{target} as |
| 2739 | |
| 2740 | @itemize @bullet |
| 2741 | @item |
| 2742 | a generic function (the most common case) |
| 2743 | |
| 2744 | @item |
| 2745 | a procedure |
| 2746 | |
| 2747 | @item |
| 2748 | a primitive generic (@pxref{Extending Guiles Primitives}) |
| 2749 | @end itemize |
| 2750 | |
| 2751 | By defining further methods for @code{add-method!}, you can |
| 2752 | theoretically handle adding methods to further types of target. |
| 2753 | @end itemize |
| 2754 | |
| 2755 | @node Generic Function Invocation |
| 2756 | @subsection Generic Function Invocation |
| 2757 | |
| 2758 | [ *fixme* Description required here. ] |
| 2759 | |
| 2760 | @code{apply-generic} |
| 2761 | |
| 2762 | @itemize @bullet |
| 2763 | @item |
| 2764 | @code{no-method} |
| 2765 | |
| 2766 | @item |
| 2767 | @code{compute-applicable-methods} |
| 2768 | |
| 2769 | @item |
| 2770 | @code{sort-applicable-methods} |
| 2771 | |
| 2772 | @item |
| 2773 | @code{apply-methods} |
| 2774 | |
| 2775 | @item |
| 2776 | @code{no-applicable-method} |
| 2777 | @end itemize |
| 2778 | |
| 2779 | @code{sort-applicable-methods} |
| 2780 | |
| 2781 | @itemize @bullet |
| 2782 | @item |
| 2783 | @code{method-more-specific?} |
| 2784 | @end itemize |
| 2785 | |
| 2786 | @code{apply-methods} |
| 2787 | |
| 2788 | @itemize @bullet |
| 2789 | @item |
| 2790 | @code{apply-method} |
| 2791 | @end itemize |
| 2792 | |
| 2793 | @code{next-method} |
| 2794 | |
| 2795 | @itemize @bullet |
| 2796 | @item |
| 2797 | @code{no-next-method} |
| 2798 | @end itemize |
| 2799 | |
| 2800 | @node Tutorial |
| 2801 | @section Tutorial |
| 2802 | @include goops-tutorial.texi |