[clinton/website/site/unknownlamer.org.git] / Metaobject Protocols.html

<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
    "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
  <head>
    <title>Metaobject Protocols</title>
    <meta name="generator" content="muse.el" />
    <meta http-equiv="Content-Type"
          content="text/html; charset=utf-8" />
    <meta name="viewport"
	  content="width=device-width, initial-scale=1.0" />
    <link href="https://feeds.unknownlamer.org/rss/site-updates" 
          rel="alternate" type="application/rss+xml" title="Updates Feed" />

<link rel="stylesheet" href="default.css" />
  </head>
  <body>
    <h1>Metaobject Protocols</h1>
   <div class="contents">
<dl>
<dt>
<a href="#sec1">Background</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec2">Object Protocols</a>
</dt>
<dt>
<a href="#sec3">CLOS Way of OO</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec4">Classes for Scratch Data and Types</a>
</dt>
<dt>
<a href="#sec5">Generics with Methods that Implement Protocols</a>
</dt>
</dl>
</dd>
</dl>
</dd>
<dt>
<a href="#sec6">Limitations of Default Language Behavior</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec7">Slot Storage</a>
</dt>
<dt>
<a href="#sec8">Design Patterns</a>
</dt>
</dl>
</dd>
<dt>
<a href="#sec9">Metasoftware</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec10">Runtime Generated Classes</a>
</dt>
<dt>
<a href="#sec11">Object Inspection</a>
</dt>
</dl>
</dd>
<dt>
<a href="#sec12">Metaobject Protocols</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec13">Limited/Generalized Internals of the Implementation</a>
</dt>
<dt>
<a href="#sec14">Classes of MOPs</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec15">Reflective</a>
</dt>
<dt>
<a href="#sec16">Intercessory</a>
</dt>
</dl>
</dd>
<dt>
<a href="#sec17">Violation of Encapsulation?</a>
</dt>
</dl>
</dd>
<dt>
<a href="#sec18">MOP Design Principles</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec19">Layered Protocol</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec20">Top Level <strong>Must</strong> Call Lower Level Methods</a>
</dt>
<dt>
<a href="#sec21">Lower Level Methods are Easier to Customize</a>
</dt>
</dl>
</dd>
<dt>
<a href="#sec22">Functional Where Possible</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec23">Memoization</a>
</dt>
<dt>
<a href="#sec24">Constant Shared Return Values</a>
</dt>
</dl>
</dd>
<dt>
<a href="#sec25">Procedural Only Where Necessary</a>
</dt>
<dt>
<a href="#sec26">Real World</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec27">UCW and Arnesi</a>
</dt>
<dt>
<a href="#sec28">CLSQL</a>
</dt>
<dt>
<a href="#sec29">Elephant</a>
</dt>
</dl>
</dd>
</dl>
</dd>
<dt>
<a href="#sec30">Sources and Further Reading</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec31">Sources</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec32">The Art of the Metaobject Protocol</a>
</dt>
<dt>
<a href="#sec33">CLOS MOP Specification</a>
</dt>
<dt>
<a href="#sec34">Metaobject Protocols: Why We Want Them and What Else They Can Do</a>
</dt>
<dt>
<a href="#sec35">Why Are Black Boxes so Hard to Reuse?</a>
</dt>
</dl>
</dd>
<dt>
<a href="#sec36">Further Reading</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec37">A Metaobject Protocol for C++</a>
</dt>
<dt>
<a href="#sec38">Open Implementations and Metaobject Protocols</a>
</dt>
</dl>
</dd>
<dt>
<a href="#sec39">Software</a>
</dt>
<dd>
<dl>
<dt>
<a href="#sec40">Closer to MOP</a>
</dt>
</dl>
</dd>
</dl>
</dd>
</dl>
</div>


<!-- Page published by Emacs Muse begins here -->
<p>In Fall of 2006 I did a small project on Metaobject Protocols for my
CS 331 class. Here lie my notes which may perhaps be useful to
others. I hope to expand them into something more useful over time.</p>

<h2><a name="sec1" id="sec1"></a>
Background</h2>

<h3><a name="sec2" id="sec2"></a>
Object Protocols</h3>

<p class="first">An object protocol is a set of methods and specification of the
interactions between the methods which provide some generic behavior
(e.g. of a sequence) that are then implemented by classes which
conform to the protocol (e.g. a vector or list). In most object
systems a class contains both the methods which implement a protocol
and the data used by the implementation. The intent is to emulate
state machines which pass messages between each other.</p>


<h3><a name="sec3" id="sec3"></a>
CLOS Way of OO</h3>

<p class="first">The Common Lisp Object System (CLOS) is different. It separates
the data and method concepts into classes and generics. A class
contains data fields only, and a generic has methods specialized for
certain types attached to it. This seems a bit weird at first, but is
significantly more powerful as it encourages complete encapsulation
through its use of classes primarily for method specialization rather
than for state storage.</p>

<h4><a name="sec4" id="sec4"></a>
Classes for Scratch Data and Types</h4>

<p class="first">In CLOS classes store data in slots (which are the same as data
members). Encapsulation is not provided; any bit of code can use
<code>slot-value</code> to access or set the value of a slot. This may seem odd at
first, but encapsulation is of questionable importance as the slots
are meant only to be used by the protocol defined around the class.</p>

<p>Classes are defined with <code>defclass</code></p>

<pre class="src">
(<span class="emacs-face-keyword">defclass</span> <span class="emacs-face-type">name</span> (superclasses ...)
  ((slot-name <span class="emacs-face-builtin">:accessor</span> slot-accessor ...)
   ...)
  (class-options ...))

(<span class="emacs-face-keyword">defclass</span> <span class="emacs-face-type">example</span> ()
  ((foo <span class="emacs-face-builtin">:accessor</span> foo-of <span class="emacs-face-builtin">:initform</span> 5)))

(<span class="emacs-face-keyword">defclass</span> <span class="emacs-face-type">example-child</span> (example)
  ((bar <span class="emacs-face-builtin">:accessor</span> bar-of <span class="emacs-face-builtin">:initform</span> (list 1 2 3))))
</pre>

<p>Slot definitions have several options; the above example shows only the
<code>:accessor</code> and <code>:initform</code> options which are the most commonly
used. <code>:accessor</code> generates an accessor for the slot (e.g. if you have
an instance of <code>example</code> you can <code>(setf (foo-of some-example-instance)
'some-value)</code> to set and <code>(foo-of some-example-instance)</code> to access the
value). <code>:initform</code> provides a default initial value for the slot as a
symbolic expression to be evaluated when an instance is created in the
lexical environment of the class definition.</p>


<h4><a name="sec5" id="sec5"></a>
Generics with Methods that Implement Protocols</h4>

<p class="first">Generics are like normal functions in Lisp, but they only provide a
lambda list (parameter list). Methods are added to the generic which
specialize on the types of their parameters and provide an
implementation. This allows writing rich layered protocols which can
enable selective modification of individual facets with minimal code.</p>

<pre class="src">
(<span class="emacs-face-keyword">defgeneric</span> <span class="emacs-face-function-name">generic</span> (parameters ...)
  (options) ...)

(<span class="emacs-face-keyword">defmethod</span> <span class="emacs-face-function-name">generic-name</span> ((parameter type) parameter ...)
  <span class="emacs-face-doc">"documentation string"</span>
  body)

(<span class="emacs-face-keyword">defgeneric</span> <span class="emacs-face-function-name">foo</span> (bar baz quux)
  (<span class="emacs-face-builtin">:documentation</span> <span class="emacs-face-doc">"Process the baz with the quux capacitor to make the
foo widget fly into the sky at warp speed"</span>))

(<span class="emacs-face-keyword">defmethod</span> <span class="emacs-face-function-name">foo</span> ((bar example) baz (quux capacitor))
  (launch bar (process-with quux baz)))
</pre>

<p>A method lambda list differs from a normal lambda list only in that it
can specify the type of the parameter using the notation <code>(name type)</code>.
Note also that methods can specialize on the types of every
argument and not just the first one. This is quite powerful for
reasons outside of the scope of this presentation.</p>


<h2><a name="sec6" id="sec6"></a>
Limitations of Default Language Behavior</h2>

<p class="first">The behavior of a language is a compromise between many competing
issues that attempts to be as generally useful as possible so that
<em>most</em> applications will have no issue with the default behavior. There
are, however, certain applications that could be cleanly written with
minor modifications to the behavior of the language, but would be
impossible or quite difficult to write otherwise.</p>

<h3><a name="sec7" id="sec7"></a>
Slot Storage</h3>

<p class="first">Most languages choose to preallocate storage for all of the slots of
an instance. Now imagine a contact database that stores information
about people in slots of a class. There may be dozens of slots, but
often many of them will be left blank. If slot storage is preallocated
much memory will be wasted and the database may not be able to fit
into the memory of the hardware it must run on (perhaps for financial
reasons, huge datasets, etc.).</p>

<p>To save memory the author of the contact database must implement his
own system to store properties and allocate them lazily. This
represents a fair bit of effort, and would implement a system that
differed from the existing slot system of classes only regarding slot
storage.</p>

<p>It would be useful if there were a way to customize slot allocation in
instances. The customizations would be minor and require overriding
only the initial allocation behavior and the behavior of the first
assignment to the slot. It is a a trivial problem in a language that
allows customization of these behaviors.</p>


<h3><a name="sec8" id="sec8"></a>
Design Patterns</h3>

<p class="first">Design Patterns are generalized versions of common patterns found in
programs. Many of them are merely methods to get around deficiencies
in the language, and can be quite messy to implement in some
languages. Ideally a pattern would be subsumed by the language, but
real world constraints require language standards to remain fairly
static.</p>


<h2><a name="sec9" id="sec9"></a>
Metasoftware</h2>

<p class="first">Some types of programs could be written easily if the language were
customizable but are nearly impossible to write when it is not.</p>

<h3><a name="sec10" id="sec10"></a>
Runtime Generated Classes</h3>

<p class="first">Say you wanted to write a video game where players could create their
own objects, attach behaviors to the objects, and perhaps mix
different objects together to create new ones. When you abstract the
problem this looks just like an object system!  Wouldn't it be nice if
your program could create new classes and methods on the fly portably?</p>


<h3><a name="sec11" id="sec11"></a>
Object Inspection</h3>

<p class="first">Imagine you were developing a complicated program with many different
objects that interacted in fairly complex ways. A tool to inspect the
structure of objects while debugging would be quite useful, but in a
traditional language would be impossible to implement portably. This
could force you to purchase a certain compiler implementation which
provided an inspector, and even then would likely not be customizable.</p>

<p>This problem can be generalized to apply to most debugging tools; it
would be useful to write such tools portably because users of the
<em>language</em> and not the <em>compiler</em> need to debug software. Sharing
infrastructure would result in better tools (more developers), and
save the man-years of wasted effort that comes with having to rewrite
unportable tools from scratch multiple times.</p>


<h2><a name="sec12" id="sec12"></a>
Metaobject Protocols</h2>

<h3><a name="sec13" id="sec13"></a>
Limited/Generalized Internals of the Implementation</h3>

<p class="first">A Metaobject Protocol (MOP) is a generalized and limited subset of the
underlying language implementation. It is limited to allow multiple
implementation strategies; this, along with careful design, is
essential because programming language research is ever advancing and
new techniques for creating more reliable and faster implementations
are still being discovered.</p>

<p>This subset of the implementation is exported as a set of methods on
metaobjects. Thus the language is implemented in itself. The system
can then be customized using the extension and overriding features of
the language itself.</p>


<h3><a name="sec14" id="sec14"></a>
Classes of MOPs</h3>

<h4><a name="sec15" id="sec15"></a>
Reflective</h4>

<p class="first">A reflective MOP provides an interface to information <em>about</em> the
running system. It exposes class relationships, the methods attached
to a generic, etc. A reflective MOP often provides some functionality
for creating new classes at runtime. Smalltalk was one of the first
languages to expose a reflective MOP.</p>

<h5>Example: Object Inspector</h5>

<pre class="src">
(<span class="emacs-face-keyword">defgeneric</span> <span class="emacs-face-function-name">example-inspect</span> (instance)
  (<span class="emacs-face-builtin">:documentation</span> <span class="emacs-face-doc">"Simple object inspector using CLOS MOP"</span>))

(<span class="emacs-face-keyword">defmethod</span> <span class="emacs-face-function-name">example-inspect</span> ((instance t))
  (format t <span class="emacs-face-string">"Simple Object~% Value: ~S~%"</span> instance))

(<span class="emacs-face-keyword">defmethod</span> <span class="emacs-face-function-name">example-inspect</span> ((instance standard-object))
  (<span class="emacs-face-keyword">let</span> ((class (class-of instance)))
    (format t <span class="emacs-face-string">"Class: ~S, Superclasses: ~S~%"</span>
            (class-name class)
            (mapcar #'class-name
                    (class-precedence-list class)))
    (<span class="emacs-face-keyword">let</span> ((slot-names (mapcar #'slot-definition-name
                              (class-slots class))))
      (format t <span class="emacs-face-string">"Slots: ~%~{ ~S~%~}"</span> slot-names)
      (inspect-loop slot-names instance #'example-inspect))))

(<span class="emacs-face-keyword">defun</span> <span class="emacs-face-function-name">inspect-loop</span> (slots instance inspector)
  (format t <span class="emacs-face-string">"Enter slot to inspect or :pop to go up one level: "</span>)
  (finish-output)
  (<span class="emacs-face-keyword">let*</span> ((slot (read))
         (found-slot (member slot slots)))
    (<span class="emacs-face-keyword">cond</span> (found-slot
           (funcall inspector (slot-value instance slot))
           (funcall inspector instance))
          ((eq slot <span class="emacs-face-builtin">:pop</span>) t)
          (t
           (format t <span class="emacs-face-string">"~S is invalid. Valid slot names: ~S~%"</span>
                   slot
                   slots)
           (inspect-loop slots instance inspector)))))
</pre>


<h5>Example: Runtime Generated Classes and Methods</h5>


<h4><a name="sec16" id="sec16"></a>
Intercessory</h4>

<p class="first">Intercessory MOPs allow the user to customize language behavior by
implementing methods which override certain aspects of the language
behavior. This class of MOPs are what make MOPs especially
powerful. No longer must a problem be restructured to fit the
implementation language; the underlying language can be reshaped to
fit the task at hand, and obfuscation of the intended structure of the
application can be avoided.</p>

<h5>Example: Lazily Allocated Slots</h5>


<h5>Example: Observer Design Pattern</h5>

<p>A simple implementation of the observer pattern is under 100 lines,
and the user level code requires only a single line of code to make
any existing class observable.</p>

<p>In a language lacking a MOP, implementing the observer pattern
requires modifying every accessor of a class to explicitly invoke any
observers, and necessitates the addition of a mixin class to the class
hierarchy. The fact that an object can be observed is a meta property
of the class, and forcing it to be implemented at the application
level dirties the inheritance hierarchy and adds unnecessary meta
details to the program.</p>

<pre class="src">
<span class="emacs-face-comment-delimiter">;;; </span><span class="emacs-face-comment">This metaclass adds a slot to instances which use it, and so the
</span><span class="emacs-face-comment-delimiter">;;; </span><span class="emacs-face-comment">system is defined in its own package to avoid name conflicts
</span>(<span class="emacs-face-keyword">defpackage</span> <span class="emacs-face-type">:observer</span>
  (<span class="emacs-face-builtin">:use</span> <span class="emacs-face-builtin">:cl</span> <span class="emacs-face-builtin">:c2mop</span>)
  (<span class="emacs-face-builtin">:export</span> observable register-observer unregister-observer))

(<span class="emacs-face-keyword">in-package</span> <span class="emacs-face-builtin">:observer</span>)

<span class="emacs-face-comment-delimiter">;;; </span><span class="emacs-face-comment">Metaclass
</span>(<span class="emacs-face-keyword">defclass</span> <span class="emacs-face-type">observable</span> (standard-class)
  ()
  (<span class="emacs-face-builtin">:documentation</span> <span class="emacs-face-doc">"Metaclass for observable objects"</span>))

(<span class="emacs-face-keyword">defmethod</span> <span class="emacs-face-function-name">compute-slots</span> ((class observable))
  <span class="emacs-face-doc">"Add a slot for storing observers to observable instances"</span>
  (cons (make-instance 'standard-effective-slot-definition
                       <span class="emacs-face-builtin">:name</span> 'observers
                       <span class="emacs-face-builtin">:initform</span> '(make-hash-table)
                       <span class="emacs-face-builtin">:initfunction</span> #'(<span class="emacs-face-keyword">lambda</span> () (make-hash-table)))
        (call-next-method)))

(<span class="emacs-face-keyword">defmethod</span> <span class="emacs-face-function-name">validate-superclass</span> ((class observable)
                                (super standard-class))
  t)

(<span class="emacs-face-keyword">defun</span> <span class="emacs-face-function-name">register-observer</span> (instance slot-name key closure)
  (register-observer-with-class (class-of instance)
                                instance
                                slot-name
                                key
                                closure))

(<span class="emacs-face-keyword">defun</span> <span class="emacs-face-function-name">unregister-observer</span> (instance slot-name key)
  (unregister-observer-with-class (class-of instance)
                                  instance
                                  slot-name
                                  key))

(<span class="emacs-face-keyword">defun</span> <span class="emacs-face-function-name">get-observers</span> (instance slot-name)
  (get-observers-with-class (class-of instance)
                            instance
                            slot-name))

(<span class="emacs-face-keyword">defun</span> <span class="emacs-face-function-name">add-observer-table</span> (instance slot-name)
  (setf (gethash slot-name (slot-value instance
                                       'observers))
        (make-hash-table)))

(<span class="emacs-face-keyword">defgeneric</span> <span class="emacs-face-function-name">register-observer-with-class</span> (class instance slot-name key closure))
(<span class="emacs-face-keyword">defgeneric</span> <span class="emacs-face-function-name">unregister-observer-with-class</span> (class
                                            instance
                                            slot-name
                                            key))

(<span class="emacs-face-keyword">defmethod</span> <span class="emacs-face-function-name">register-observer-with-class</span> ((class observable)
                                         instance
                                         slot-name
                                         key
                                         closure)
  (setf (gethash key
                 (or (gethash slot-name
                              (slot-value instance 'observers))
                     <span class="emacs-face-comment-delimiter">;; </span><span class="emacs-face-comment">Lazily add observer hash tables
</span>                     (add-observer-table instance slot-name)))
        closure))

(<span class="emacs-face-keyword">defmethod</span> <span class="emacs-face-function-name">unregister-observer-with-class</span> ((class observable)
                                           instance
                                           slot-name
                                           key)
  (remhash key (gethash slot-name
                        (slot-value instance 'observers))))

(<span class="emacs-face-keyword">defmethod</span> <span class="emacs-face-function-name">get-observers-with-class</span> ((class observable)
                                     instance
                                     slot-name)
  (gethash slot-name (slot-value instance 'observers)))

(<span class="emacs-face-keyword">defmethod</span> (<span class="emacs-face-function-name">setf slot-value-using-class</span>) <span class="emacs-face-builtin">:before</span> (new-value
                                                  (class observable)
                                                  instance
                                                  slot)
  (<span class="emacs-face-keyword">let</span> ((slot-name (slot-definition-name slot)))
    (<span class="emacs-face-keyword">if</span> (not (eq 'observers slot-name))
        (<span class="emacs-face-keyword">let</span> ((observers
               (get-observers instance (slot-definition-name slot))))
          (<span class="emacs-face-keyword">if</span> observers
              (maphash #'(<span class="emacs-face-keyword">lambda</span> (key observer)
                           (funcall observer
                                    (<span class="emacs-face-keyword">if</span> (slot-boundp instance slot-name)
                                        (slot-value instance slot-name)
                                      nil)
                                    new-value))
                       observers))))))
</pre>


<h3><a name="sec17" id="sec17"></a>
Violation of Encapsulation?</h3>

<p class="first">A MOP may seem like a violation of encapsulation by revealing some
implementation details, but in reality a well designed protocol does
not reveal anything which was not already exposed. Implementation
decisions affect users, and some of these details do leak through to
higher levels (e.g. the memory layout of slots). Implicit in the
protocol specification are these implementation details, and the MOP
merely makes this limited subset available for customization.</p>

<p>A MOP makes it possible to customize certain implementation decisions
that do not <strong>radically</strong> alter the behavior of the base language. The
conceptual vocabulary of the system retains its meaning, and so code
written in one dialect can interact with code written in another
without knowing that they speak different ones.</p>


<h2><a name="sec18" id="sec18"></a>
MOP Design Principles</h2>

<h3><a name="sec19" id="sec19"></a>
Layered Protocol</h3>

<p class="first">A layered protocol design is good for both meta and normal object
protocols, and enables a combinatorial explosion of customizations to
the protocol.</p>

<h4><a name="sec20" id="sec20"></a>
Top Level <strong>Must</strong> Call Lower Level Methods</h4>

<p class="first">The top level methods of a layered protocol are required to call
certain lower level methods to perform some tasks. This both makes it
easier to customize the top level methods (which perform very broad
tasks) by providing some pieces of implementation for the programmer,
and enables more customization by opening up the replacement of lower
level functions as a way to alter a small detail of the high level
behavior.</p>


<h4><a name="sec21" id="sec21"></a>
Lower Level Methods are Easier to Customize</h4>

<p class="first">The lower level methods of a MOP are limited in scope and can be
implemented easily. Often the desired changes to language behavior are
minor, and having methods that perform simple tasks which are often
customized reduces the effort required to extend the system.</p>


<h3><a name="sec22" id="sec22"></a>
Functional Where Possible</h3>

<p class="first">Functional protocols are preferred for MOPs (and object protocols in
general). Functional protocols open up several optimizations for the
implementation without burdening the user of the protocol.</p>

<h4><a name="sec23" id="sec23"></a>
Memoization</h4>

<p class="first">Memoization is the process of saving the results of a function call
for future use. This avoids expensive recomputation of values which
have not changed (recall that a true function will always return the
same result when given the same arguments).</p>

<p>A functional MOP can be optimized easily by exploiting this property
to memoize the return values of calls to expensive operations. A MOP
must be be very fast to avoid making programs unusably slow, and
memoization is able to give an appreciable speedup in many cases
without a significant burden on memory usage.</p>


<h4><a name="sec24" id="sec24"></a>
Constant Shared Return Values</h4>

<p class="first">Disallowing modification of values returned by protocol methods allows
the implementation to return large data structures by reference to
avoid expensive copying without having to do expensive data integrity
checks or copying.</p>


<h3><a name="sec25" id="sec25"></a>
Procedural Only Where Necessary</h3>

<p class="first">Some operations like method invocation are inherently stateful and so
must use a procedural protocol. There is no benefit to be gained from
using a functional protocol, and indeed an attempt would result in
obtuse code that severely restricted the implementation. Do note that
only a very small part of method invocation is stateful (the actual
call), and most of it can be implemented functionally (e.g. computing
the discriminating function).</p>


<h3><a name="sec26" id="sec26"></a>
Real World</h3>

<h4><a name="sec27" id="sec27"></a>
<a href="http://common-lisp.net/project/ucw/">UCW</a> and <a href="http://common-lisp.net/project/bese/arnesi.html">Arnesi</a></h4>

<p class="first">Arnesi uses the CLOS MOP to implement methods which are transparently
rewritten into continuation passing style. This allows their execution
to be suspended at certain points and resumed later. UCW builds on top
of this to support a web framework where the statelessness of http is
hidden from the user; displaying a page suspends the execution of the
current continuation, and resumes it upon submission. The user level
code is completely unaware of this.</p>


<h4><a name="sec28" id="sec28"></a>
<a href="http://clsql.b9.com">CLSQL</a></h4>

<p class="first">CLSQL uses the reflective part of the CLOS MOP to map Common Lisp data
types into SQL types, and the intercessory protocol for slot
allocation to map slots onto database columns or sql expressions (for
implementing relational slots).</p>


<h4><a name="sec29" id="sec29"></a>
<a href="http://common-lisp.net/project/elephant/">Elephant</a></h4>

<p class="first">Elephant uses the CLOS MOP to transparently store any class to disk
and handle paging between the disk store and memory efficiently
without user intervention.</p>


<h2><a name="sec30" id="sec30"></a>
Sources and Further Reading</h2>

<h3><a name="sec31" id="sec31"></a>
Sources</h3>

<h4><a name="sec32" id="sec32"></a>
The Art of the Metaobject Protocol</h4>

<h5>Kiczales, Gregor et al. MIT Press 1991</h5>

<p>Highly recommended reading even if you plan to never implement a MOP
or use the CLOS one. The design principles it recommends are quite
useful.</p>


<h4><a name="sec33" id="sec33"></a>
<a href="http://www.lisp.org/mop/contents.html">CLOS MOP Specification</a></h4>

<p class="first">Specification of the MOP for CLOS defined in <em>The Art of the Metaobject Protocol</em>.</p>


<h4><a name="sec34" id="sec34"></a>
<a href="http://citeseer.ist.psu.edu/399658.html">Metaobject Protocols: Why We Want Them and What Else They Can Do</a></h4>

<p class="first">A short overview of MOP design principles followed by three example
metaobject protocols for Scheme.</p>


<h4><a name="sec35" id="sec35"></a>
<a href="http://www2.parc.com/csl/groups/sda/projects/oi/towards-talk/transcript.html">Why Are Black Boxes so Hard to Reuse?</a></h4>

<p class="first">Transcription of a talk on the benefits of open implementations of
software. It first discusses several problems that black box software
implementations pose, and then presents existing solutions. It shows
how the existing solutions are insufficient, and then provides
metaobject protocols as a solution to most of the problems.</p>


<h3><a name="sec36" id="sec36"></a>
Further Reading</h3>

<h4><a name="sec37" id="sec37"></a>
<a href="http://citeseer.ist.psu.edu/chiba95metaobject.html">A Metaobject Protocol for C++</a></h4>

<p class="first">Example of a purely compile time MOP. It implements the functionality
of a code walker and something similar to the Lisp macro system.</p>


<h4><a name="sec38" id="sec38"></a>
<a href="http://www.parc.com/csl/groups/sda/publications/papers/Kiczales-TUT95/for-web.pdf">Open Implementations and Metaobject Protocols</a></h4>

<p class="first">It is a bit long, but it seems to follow a similar structure to AMOP
in introducing MOPs and their usefulness. The pages are slides with
notes, and so the 331 pages might not actually take that long to read.</p>


<h3><a name="sec39" id="sec39"></a>
Software</h3>

<h4><a name="sec40" id="sec40"></a>
<a href="http://common-lisp.net/project/closer/closer-mop.html">Closer to MOP</a></h4>

<p class="first">Compatibility layer that attempts to present the <em>Art of the Metaobject
Protocol</em> MOP specification properly in as many Common Lisp
implementation as possible.</p>


  <!-- Page published by Emacs Muse ends here -->

  <p class="cke-buttons">
    <!-- validating badges, any browser, etc -->
    <a href="https://validator.w3.org/check/referer"><img
	src="https://www.w3.org/Icons/valid-xhtml10"
	alt="Valid XHTML 1.0!" /></a>
    
    <a href="https://www.anybrowser.org/campaign/"><img
	src="img/buttons/w3c_ab.png" alt="[ Viewable With Any Browser
	]" /></a>

    <a href="https://www.debian.org/"><img
	src="img/buttons/debian.png" alt="[ Powered by Debian ]" /></a>
    
    <a href="https://hcoop.net/">
      <img src="img/buttons/hcoop.png" 
	alt="[ Hosted by HCoop]" />
    </a>

    <a href="https://www.fsf.org/register_form?referrer=114">
      <img src="img/buttons/fsf_member.png"
	alt="[ FSF Associate Member ]" />
    </a>
  </p>

<p class="cke-footer">&lt;ascii_phil&gt; There once was a man named Bertold
&lt;ascii_phil&gt; Who drank beer when the weather grew cold
&lt;ascii_phil&gt; As he reached for his cup...
&lt;ascii_phil&gt; "NEEEEVER GONNA GIVE YOU UP!!!"
&lt;ascii_phil&gt; Oh, snap! You just got limerickrolled!
</p>
<p class="cke-timestamp">Last Modified:
    January 21, 2013</p>
  </body>
</html>