Import Upstream version 20180207
[hcoop/debian/mlton.git] / doc / guide / localhost / PolymorphicEquality
1 <!DOCTYPE html>
2 <html lang="en">
3 <head>
4 <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
5 <meta name="generator" content="AsciiDoc 8.6.9">
6 <title>PolymorphicEquality</title>
7 <link rel="stylesheet" href="./asciidoc.css" type="text/css">
8 <link rel="stylesheet" href="./pygments.css" type="text/css">
9
10
11 <script type="text/javascript" src="./asciidoc.js"></script>
12 <script type="text/javascript">
13 /*<![CDATA[*/
14 asciidoc.install(2);
15 /*]]>*/
16 </script>
17 <link rel="stylesheet" href="./mlton.css" type="text/css">
18 </head>
19 <body class="article">
20 <div id="banner">
21 <div id="banner-home">
22 <a href="./Home">MLton 20180207</a>
23 </div>
24 </div>
25 <div id="header">
26 <h1>PolymorphicEquality</h1>
27 <div id="toc">
28 <div id="toctitle">Table of Contents</div>
29 <noscript><p><b>JavaScript must be enabled in your browser to display the table of contents.</b></p></noscript>
30 </div>
31 </div>
32 <div id="content">
33 <div id="preamble">
34 <div class="sectionbody">
35 <div class="paragraph"><p>Polymorphic equality is a built-in function in
36 <a href="StandardML">Standard ML</a> that compares two values of the same type
37 for equality. It is specified as</p></div>
38 <div class="listingblock">
39 <div class="content"><div class="highlight"><pre><span class="k">val</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="p">:</span><span class="w"> </span><span class="n">&#39;&#39;a</span><span class="w"> </span><span class="n">*</span><span class="w"> </span><span class="n">&#39;&#39;a</span><span class="w"> </span><span class="p">-&gt;</span><span class="w"> </span><span class="n">bool</span><span class="w"></span>
40 </pre></div></div></div>
41 <div class="paragraph"><p>The <span class="monospaced">''a</span> in the specification are
42 <a href="EqualityTypeVariable">equality type variables</a>, and indicate that
43 polymorphic equality can only be applied to values of an
44 <a href="EqualityType">equality type</a>. It is not allowed in SML to rebind
45 <span class="monospaced">=</span>, so a programmer is guaranteed that <span class="monospaced">=</span> always denotes polymorphic
46 equality.</p></div>
47 </div>
48 </div>
49 <div class="sect1">
50 <h2 id="_equality_of_ground_types">Equality of ground types</h2>
51 <div class="sectionbody">
52 <div class="paragraph"><p>Ground types like <span class="monospaced">char</span>, <span class="monospaced">int</span>, and <span class="monospaced">word</span> may be compared (to values
53 of the same type). For example, <span class="monospaced">13 = 14</span> is type correct and yields
54 <span class="monospaced">false</span>.</p></div>
55 </div>
56 </div>
57 <div class="sect1">
58 <h2 id="_equality_of_reals">Equality of reals</h2>
59 <div class="sectionbody">
60 <div class="paragraph"><p>The one ground type that can not be compared is <span class="monospaced">real</span>. So,
61 <span class="monospaced">13.0 = 14.0</span> is not type correct. One can use <span class="monospaced">Real.==</span> to compare
62 reals for equality, but beware that this has different algebraic
63 properties than polymorphic equality.</p></div>
64 <div class="paragraph"><p>See <a href="http://standardml.org/Basis/real.html">http://standardml.org/Basis/real.html</a> for a discussion of why
65 <span class="monospaced">real</span> is not an equality type.</p></div>
66 </div>
67 </div>
68 <div class="sect1">
69 <h2 id="_equality_of_functions">Equality of functions</h2>
70 <div class="sectionbody">
71 <div class="paragraph"><p>Comparison of functions is not allowed.</p></div>
72 </div>
73 </div>
74 <div class="sect1">
75 <h2 id="_equality_of_immutable_types">Equality of immutable types</h2>
76 <div class="sectionbody">
77 <div class="paragraph"><p>Polymorphic equality can be used on <a href="Immutable">immutable</a> values like
78 tuples, records, lists, and vectors. For example,</p></div>
79 <div class="listingblock">
80 <div class="content monospaced">
81 <pre>(1, 2, 3) = (4, 5, 6)</pre>
82 </div></div>
83 <div class="paragraph"><p>is a type-correct expression yielding <span class="monospaced">false</span>, while</p></div>
84 <div class="listingblock">
85 <div class="content monospaced">
86 <pre>[1, 2, 3] = [1, 2, 3]</pre>
87 </div></div>
88 <div class="paragraph"><p>is type correct and yields <span class="monospaced">true</span>.</p></div>
89 <div class="paragraph"><p>Equality on immutable values is computed by structure, which means
90 that values are compared by recursively descending the data structure
91 until ground types are reached, at which point the ground types are
92 compared with primitive equality tests (like comparison of
93 characters). So, the expression</p></div>
94 <div class="listingblock">
95 <div class="content monospaced">
96 <pre>[1, 2, 3] = [1, 1 + 1, 1 + 1 + 1]</pre>
97 </div></div>
98 <div class="paragraph"><p>is guaranteed to yield <span class="monospaced">true</span>, even though the lists may occupy
99 different locations in memory.</p></div>
100 <div class="paragraph"><p>Because of structural equality, immutable values can only be compared
101 if their components can be compared. For example, <span class="monospaced">[1, 2, 3]</span> can be
102 compared, but <span class="monospaced">[1.0, 2.0, 3.0]</span> can not. The SML type system uses
103 <a href="EqualityType">equality types</a> to ensure that structural equality is
104 only applied to valid values.</p></div>
105 </div>
106 </div>
107 <div class="sect1">
108 <h2 id="_equality_of_mutable_values">Equality of mutable values</h2>
109 <div class="sectionbody">
110 <div class="paragraph"><p>In contrast to immutable values, polymorphic equality of
111 <a href="Mutable">mutable</a> values (like ref cells and arrays) is performed by
112 pointer comparison, not by structure. So, the expression</p></div>
113 <div class="listingblock">
114 <div class="content monospaced">
115 <pre>ref 13 = ref 13</pre>
116 </div></div>
117 <div class="paragraph"><p>is guaranteed to yield <span class="monospaced">false</span>, even though the ref cells hold the
118 same contents.</p></div>
119 <div class="paragraph"><p>Because equality of mutable values is not structural, arrays and refs
120 can be compared <em>even if their components are not equality types</em>.
121 Hence, the following expression is type correct (and yields true).</p></div>
122 <div class="listingblock">
123 <div class="content"><div class="highlight"><pre><span class="k">let</span><span class="w"></span>
124 <span class="w"> </span><span class="k">val</span><span class="w"> </span><span class="n">r</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">ref</span><span class="w"> </span><span class="mf">13.0</span><span class="w"></span>
125 <span class="k">in</span><span class="w"></span>
126 <span class="w"> </span><span class="n">r</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">r</span><span class="w"></span>
127 <span class="k">end</span><span class="w"></span>
128 </pre></div></div></div>
129 </div>
130 </div>
131 <div class="sect1">
132 <h2 id="_equality_of_datatypes">Equality of datatypes</h2>
133 <div class="sectionbody">
134 <div class="paragraph"><p>Polymorphic equality of datatypes is structural. Two values of the
135 same datatype are equal if they are of the same <a href="Variant">variant</a> and
136 if the <a href="Variant">variant</a>'s arguments are equal (recursively). So,
137 with the datatype</p></div>
138 <div class="listingblock">
139 <div class="content"><div class="highlight"><pre><span class="k">datatype</span><span class="w"> </span><span class="n">t</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">A</span><span class="w"> </span><span class="p">|</span><span class="w"> </span><span class="n">B</span><span class="w"> </span><span class="k">of</span><span class="w"> </span><span class="n">t</span><span class="w"></span>
140 </pre></div></div></div>
141 <div class="paragraph"><p>then <span class="monospaced">B (B A) = B A</span> is type correct and yields <span class="monospaced">false</span>, while <span class="monospaced">A = A</span>
142 and <span class="monospaced">B A = B A</span> yield <span class="monospaced">true</span>.</p></div>
143 <div class="paragraph"><p>As polymorphic equality descends two values to compare them, it uses
144 pointer equality whenever it reaches a mutable value. So, with the
145 datatype</p></div>
146 <div class="listingblock">
147 <div class="content"><div class="highlight"><pre><span class="k">datatype</span><span class="w"> </span><span class="n">t</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">A</span><span class="w"> </span><span class="k">of</span><span class="w"> </span><span class="n">int</span><span class="w"> </span><span class="n">ref</span><span class="w"> </span><span class="p">|</span><span class="w"> </span><span class="p">...</span><span class="w"></span>
148 </pre></div></div></div>
149 <div class="paragraph"><p>then <span class="monospaced">A (ref 13) = A (ref 13)</span> is type correct and yields <span class="monospaced">false</span>,
150 because the pointer equality on the two ref cells yields <span class="monospaced">false</span>.</p></div>
151 <div class="paragraph"><p>One weakness of the SML type system is that datatypes do not inherit
152 the special property of the <span class="monospaced">ref</span> and <span class="monospaced">array</span> type constructors that
153 allows them to be compared regardless of their component type. For
154 example, after declaring</p></div>
155 <div class="listingblock">
156 <div class="content"><div class="highlight"><pre><span class="k">datatype</span><span class="w"> </span><span class="n">&#39;a</span><span class="w"> </span><span class="n">t</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">A</span><span class="w"> </span><span class="k">of</span><span class="w"> </span><span class="n">&#39;a</span><span class="w"> </span><span class="n">ref</span><span class="w"></span>
157 </pre></div></div></div>
158 <div class="paragraph"><p>one might expect to be able to compare two values of type <span class="monospaced">real t</span>,
159 because pointer comparison on a ref cell would suffice.
160 Unfortunately, the type system can only express that a user-defined
161 datatype <a href="AdmitsEquality">admits equality</a> or not. In this case, <span class="monospaced">t</span>
162 admits equality, which means that <span class="monospaced">int t</span> can be compared but that
163 <span class="monospaced">real t</span> can not. We can confirm this with the program</p></div>
164 <div class="listingblock">
165 <div class="content"><div class="highlight"><pre><span class="k">datatype</span><span class="w"> </span><span class="n">&#39;a</span><span class="w"> </span><span class="n">t</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">A</span><span class="w"> </span><span class="k">of</span><span class="w"> </span><span class="n">&#39;a</span><span class="w"> </span><span class="n">ref</span><span class="w"></span>
166 <span class="k">fun</span><span class="w"> </span><span class="n">f</span><span class="w"> </span><span class="p">(</span><span class="n">x</span><span class="p">:</span><span class="w"> </span><span class="n">real</span><span class="w"> </span><span class="n">t</span><span class="p">,</span><span class="w"> </span><span class="n">y</span><span class="p">:</span><span class="w"> </span><span class="n">real</span><span class="w"> </span><span class="n">t</span><span class="p">)</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">x</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">y</span><span class="w"></span>
167 </pre></div></div></div>
168 <div class="paragraph"><p>on which MLton reports the following error.</p></div>
169 <div class="listingblock">
170 <div class="content monospaced">
171 <pre>Error: z.sml 2.32-2.36.
172 Function applied to incorrect argument.
173 expects: [&lt;equality&gt;] t * [&lt;equality&gt;] t
174 but got: [real] t * [real] t
175 in: = (x, y)</pre>
176 </div></div>
177 </div>
178 </div>
179 <div class="sect1">
180 <h2 id="_implementation">Implementation</h2>
181 <div class="sectionbody">
182 <div class="paragraph"><p>Polymorphic equality is implemented by recursively descending the two
183 values being compared, stopping as soon as they are determined to be
184 unequal, or exploring the entire values to determine that they are
185 equal. Hence, polymorphic equality can take time proportional to the
186 size of the smaller value.</p></div>
187 <div class="paragraph"><p>MLton uses some optimizations to improve performance.</p></div>
188 <div class="ulist"><ul>
189 <li>
190 <p>
191 When computing structural equality, first do a pointer comparison.
192 If the comparison yields <span class="monospaced">true</span>, then stop and return <span class="monospaced">true</span>, since
193 the structural comparison is guaranteed to do so. If the pointer
194 comparison fails, then recursively descend the values.
195 </p>
196 </li>
197 <li>
198 <p>
199 If a datatype is an enum (e.g. <span class="monospaced">datatype t = A | B | C</span>), then a
200 single comparison suffices to compare values of the datatype. No case
201 dispatch is required to determine whether the two values are of the
202 same <a href="Variant">variant</a>.
203 </p>
204 </li>
205 <li>
206 <p>
207 When comparing a known constant non-value-carrying
208 <a href="Variant">variant</a>, use a single comparison. For example, the
209 following code will compile into a single comparison for <span class="monospaced">A = x</span>.
210 </p>
211 <div class="listingblock">
212 <div class="content"><div class="highlight"><pre><span class="k">datatype</span><span class="w"> </span><span class="n">t</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">A</span><span class="w"> </span><span class="p">|</span><span class="w"> </span><span class="n">B</span><span class="w"> </span><span class="p">|</span><span class="w"> </span><span class="n">C</span><span class="w"> </span><span class="k">of</span><span class="w"> </span><span class="p">...</span><span class="w"></span>
213 <span class="k">fun</span><span class="w"> </span><span class="n">f</span><span class="w"> </span><span class="n">x</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="p">...</span><span class="w"> </span><span class="k">if</span><span class="w"> </span><span class="n">A</span><span class="w"> </span><span class="p">=</span><span class="w"> </span><span class="n">x</span><span class="w"> </span><span class="k">then</span><span class="w"> </span><span class="p">...</span><span class="w"></span>
214 </pre></div></div></div>
215 </li>
216 <li>
217 <p>
218 When comparing a small constant <span class="monospaced">IntInf.int</span> to another
219 <span class="monospaced">IntInf.int</span>, use a single comparison against the constant. No case
220 dispatch is required.
221 </p>
222 </li>
223 </ul></div>
224 </div>
225 </div>
226 <div class="sect1">
227 <h2 id="_also_see">Also see</h2>
228 <div class="sectionbody">
229 <div class="ulist"><ul>
230 <li>
231 <p>
232 <a href="AdmitsEquality">AdmitsEquality</a>
233 </p>
234 </li>
235 <li>
236 <p>
237 <a href="EqualityType">EqualityType</a>
238 </p>
239 </li>
240 <li>
241 <p>
242 <a href="EqualityTypeVariable">EqualityTypeVariable</a>
243 </p>
244 </li>
245 </ul></div>
246 </div>
247 </div>
248 </div>
249 <div id="footnotes"><hr></div>
250 <div id="footer">
251 <div id="footer-text">
252 </div>
253 <div id="footer-badges">
254 </div>
255 </div>
256 </body>
257 </html>