wenzelm@12572
|
1 |
%
|
wenzelm@12572
|
2 |
\begin{isabellebody}%
|
wenzelm@12572
|
3 |
\def\isabellecontext{Records}%
|
wenzelm@12572
|
4 |
%
|
wenzelm@12572
|
5 |
\isamarkupheader{Records \label{sec:records}%
|
wenzelm@12572
|
6 |
}
|
wenzelm@12572
|
7 |
\isamarkuptrue%
|
wenzelm@12572
|
8 |
\isamarkupfalse%
|
wenzelm@12572
|
9 |
%
|
wenzelm@12572
|
10 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
11 |
\index{records|(}%
|
wenzelm@12572
|
12 |
Records are familiar from programming languages. A record of $n$
|
wenzelm@12572
|
13 |
fields is essentially an $n$-tuple, but the record's components have
|
wenzelm@12572
|
14 |
names, which can make expressions easier to read and reduces the
|
wenzelm@12572
|
15 |
risk of confusing one field for another.
|
wenzelm@12572
|
16 |
|
wenzelm@12572
|
17 |
A basic Isabelle record covers a certain set of fields, with select
|
wenzelm@12572
|
18 |
and update operations. Each field has a specified type, which may
|
wenzelm@12572
|
19 |
be polymorphic. The field names are part of the record type, and
|
wenzelm@12572
|
20 |
the order of the fields is significant --- as it is in Pascal but
|
wenzelm@12572
|
21 |
not in Standard ML. If two different record types have field names
|
wenzelm@12572
|
22 |
in common, then the ambiguity is resolved in the usual way, by
|
wenzelm@12572
|
23 |
qualified names.
|
wenzelm@12572
|
24 |
|
wenzelm@12572
|
25 |
Record types can also be defined by extending other record types.
|
wenzelm@12572
|
26 |
Extensible records make use of the reserved pseudo-field \cdx{more},
|
wenzelm@12572
|
27 |
which is present in every record type. Generic record operations
|
wenzelm@12572
|
28 |
work on all possible extensions of a given type scheme; naive
|
wenzelm@12572
|
29 |
polymorphism takes care of structural sub-typing behind the scenes.
|
wenzelm@12572
|
30 |
There are also explicit coercion functions between fixed record
|
wenzelm@12572
|
31 |
types.%
|
wenzelm@12572
|
32 |
\end{isamarkuptext}%
|
wenzelm@12572
|
33 |
\isamarkuptrue%
|
wenzelm@12572
|
34 |
%
|
wenzelm@12572
|
35 |
\isamarkupsubsection{Record Basics%
|
wenzelm@12572
|
36 |
}
|
wenzelm@12572
|
37 |
\isamarkuptrue%
|
wenzelm@12572
|
38 |
%
|
wenzelm@12572
|
39 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
40 |
Record types are not primitive in Isabelle and have a subtle
|
wenzelm@12572
|
41 |
internal representation based on nested copies of the primitive
|
wenzelm@12572
|
42 |
product type. A \commdx{record} declaration introduces a new record
|
wenzelm@12572
|
43 |
type scheme by specifying its fields, which are packaged internally
|
wenzelm@12572
|
44 |
to hold up the perception of records as a separate concept.%
|
wenzelm@12572
|
45 |
\end{isamarkuptext}%
|
wenzelm@12572
|
46 |
\isamarkuptrue%
|
wenzelm@12572
|
47 |
\isacommand{record}\ point\ {\isacharequal}\isanewline
|
wenzelm@12572
|
48 |
\ \ Xcoord\ {\isacharcolon}{\isacharcolon}\ int\isanewline
|
wenzelm@12572
|
49 |
\ \ Ycoord\ {\isacharcolon}{\isacharcolon}\ int\isamarkupfalse%
|
wenzelm@12572
|
50 |
%
|
wenzelm@12572
|
51 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
52 |
Records of type \isa{point} have two fields named \isa{Xcoord}
|
wenzelm@12572
|
53 |
and \isa{Ycoord}, both of type~\isa{int}. We now define a
|
wenzelm@12572
|
54 |
constant of type \isa{point}:%
|
wenzelm@12572
|
55 |
\end{isamarkuptext}%
|
wenzelm@12572
|
56 |
\isamarkuptrue%
|
wenzelm@12572
|
57 |
\isacommand{constdefs}\isanewline
|
wenzelm@12572
|
58 |
\ \ pt{\isadigit{1}}\ {\isacharcolon}{\isacharcolon}\ point\isanewline
|
wenzelm@12572
|
59 |
\ \ {\isachardoublequote}pt{\isadigit{1}}\ {\isasymequiv}\ {\isacharparenleft}{\isacharbar}\ Xcoord\ {\isacharequal}\ {\isadigit{9}}{\isadigit{9}}{\isadigit{9}}{\isacharcomma}\ Ycoord\ {\isacharequal}\ {\isadigit{2}}{\isadigit{3}}\ {\isacharbar}{\isacharparenright}{\isachardoublequote}\isamarkupfalse%
|
wenzelm@12572
|
60 |
%
|
wenzelm@12572
|
61 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
62 |
We see above the ASCII notation for record brackets. You can also
|
wenzelm@12572
|
63 |
use the symbolic brackets \isa{{\isasymlparr}} and \isa{{\isasymrparr}}. Record type
|
wenzelm@12572
|
64 |
expressions can be written directly as well, without referring to
|
wenzelm@12572
|
65 |
previously declared names (which happen to be mere type
|
wenzelm@12572
|
66 |
abbreviations):%
|
wenzelm@12572
|
67 |
\end{isamarkuptext}%
|
wenzelm@12572
|
68 |
\isamarkuptrue%
|
wenzelm@12572
|
69 |
\isacommand{constdefs}\isanewline
|
wenzelm@12572
|
70 |
\ \ pt{\isadigit{2}}\ {\isacharcolon}{\isacharcolon}\ {\isachardoublequote}{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharcolon}\ int{\isacharcomma}\ Ycoord\ {\isacharcolon}{\isacharcolon}\ int{\isasymrparr}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
71 |
\ \ {\isachardoublequote}pt{\isadigit{2}}\ {\isasymequiv}\ {\isasymlparr}Xcoord\ {\isacharequal}\ {\isacharminus}{\isadigit{4}}{\isadigit{5}}{\isacharcomma}\ Ycoord\ {\isacharequal}\ {\isadigit{9}}{\isadigit{7}}{\isasymrparr}{\isachardoublequote}\isamarkupfalse%
|
wenzelm@12572
|
72 |
%
|
wenzelm@12572
|
73 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
74 |
For each field, there is a \emph{selector} function of the same
|
wenzelm@12572
|
75 |
name. For example, if \isa{p} has type \isa{point} then \isa{Xcoord\ p} denotes the value of the \isa{Xcoord} field of~\isa{p}. Expressions involving field selection of explicit records are
|
wenzelm@12572
|
76 |
simplified automatically:%
|
wenzelm@12572
|
77 |
\end{isamarkuptext}%
|
wenzelm@12572
|
78 |
\isamarkuptrue%
|
wenzelm@12572
|
79 |
\isacommand{lemma}\ {\isachardoublequote}Xcoord\ {\isasymlparr}Xcoord\ {\isacharequal}\ a{\isacharcomma}\ Ycoord\ {\isacharequal}\ b{\isasymrparr}\ {\isacharequal}\ a{\isachardoublequote}\isanewline
|
wenzelm@12572
|
80 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
81 |
\isacommand{by}\ simp\isamarkupfalse%
|
wenzelm@12572
|
82 |
%
|
wenzelm@12572
|
83 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
84 |
The \emph{update} operation is functional. For example, \isa{p{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ {\isadigit{0}}{\isasymrparr}} is a record whose \isa{Xcoord} value is zero
|
wenzelm@12572
|
85 |
and whose \isa{Ycoord} value is copied from~\isa{p}. Updates
|
wenzelm@12572
|
86 |
are also simplified automatically:%
|
wenzelm@12572
|
87 |
\end{isamarkuptext}%
|
wenzelm@12572
|
88 |
\isamarkuptrue%
|
wenzelm@12572
|
89 |
\isacommand{lemma}\ {\isachardoublequote}{\isasymlparr}Xcoord\ {\isacharequal}\ a{\isacharcomma}\ Ycoord\ {\isacharequal}\ b{\isasymrparr}{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ {\isadigit{0}}{\isasymrparr}\ {\isacharequal}\isanewline
|
wenzelm@12572
|
90 |
\ \ \ \ {\isasymlparr}Xcoord\ {\isacharequal}\ {\isadigit{0}}{\isacharcomma}\ Ycoord\ {\isacharequal}\ b{\isasymrparr}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
91 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
92 |
\isacommand{by}\ simp\isamarkupfalse%
|
wenzelm@12572
|
93 |
%
|
wenzelm@12572
|
94 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
95 |
\begin{warn}
|
wenzelm@12572
|
96 |
Field names are declared as constants and can no longer be used as
|
wenzelm@12572
|
97 |
variables. It would be unwise, for example, to call the fields of
|
wenzelm@12572
|
98 |
type \isa{point} simply \isa{x} and~\isa{y}. Each record
|
wenzelm@12572
|
99 |
declaration introduces a constant \cdx{more}.
|
wenzelm@12572
|
100 |
\end{warn}%
|
wenzelm@12572
|
101 |
\end{isamarkuptext}%
|
wenzelm@12572
|
102 |
\isamarkuptrue%
|
wenzelm@12572
|
103 |
%
|
wenzelm@12572
|
104 |
\isamarkupsubsection{Extensible Records and Generic Operations%
|
wenzelm@12572
|
105 |
}
|
wenzelm@12572
|
106 |
\isamarkuptrue%
|
wenzelm@12572
|
107 |
%
|
wenzelm@12572
|
108 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
109 |
\index{records!extensible|(}%
|
wenzelm@12572
|
110 |
|
wenzelm@12572
|
111 |
Now, let us define coloured points (type \isa{cpoint}) to be
|
wenzelm@12572
|
112 |
points extended with a field \isa{col} of type \isa{colour}:%
|
wenzelm@12572
|
113 |
\end{isamarkuptext}%
|
wenzelm@12572
|
114 |
\isamarkuptrue%
|
wenzelm@12572
|
115 |
\isacommand{datatype}\ colour\ {\isacharequal}\ Red\ {\isacharbar}\ Green\ {\isacharbar}\ Blue\isanewline
|
wenzelm@12572
|
116 |
\isanewline
|
wenzelm@12572
|
117 |
\isamarkupfalse%
|
wenzelm@12572
|
118 |
\isacommand{record}\ cpoint\ {\isacharequal}\ point\ {\isacharplus}\isanewline
|
wenzelm@12572
|
119 |
\ \ col\ {\isacharcolon}{\isacharcolon}\ colour\isamarkupfalse%
|
wenzelm@12572
|
120 |
%
|
wenzelm@12572
|
121 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
122 |
The fields of this new type are \isa{Xcoord}, \isa{Ycoord} and
|
wenzelm@12572
|
123 |
\isa{col}, in that order:%
|
wenzelm@12572
|
124 |
\end{isamarkuptext}%
|
wenzelm@12572
|
125 |
\isamarkuptrue%
|
wenzelm@12572
|
126 |
\isacommand{constdefs}\isanewline
|
wenzelm@12572
|
127 |
\ \ cpt{\isadigit{1}}\ {\isacharcolon}{\isacharcolon}\ cpoint\isanewline
|
wenzelm@12572
|
128 |
\ \ {\isachardoublequote}cpt{\isadigit{1}}\ {\isasymequiv}\ {\isasymlparr}Xcoord\ {\isacharequal}\ {\isadigit{9}}{\isadigit{9}}{\isadigit{9}}{\isacharcomma}\ Ycoord\ {\isacharequal}\ {\isadigit{2}}{\isadigit{3}}{\isacharcomma}\ col\ {\isacharequal}\ Green{\isasymrparr}{\isachardoublequote}\isamarkupfalse%
|
wenzelm@12572
|
129 |
%
|
wenzelm@12572
|
130 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
131 |
We can define generic operations that work on arbitrary instances of
|
wenzelm@12572
|
132 |
a record scheme, e.g.\ covering \isa{point}, \isa{cpoint} and any
|
wenzelm@12572
|
133 |
further extensions. Every record structure has an implicit
|
wenzelm@12572
|
134 |
pseudo-field, \cdx{more}, that keeps the extension as an explicit
|
wenzelm@12572
|
135 |
value. Its type is declared as completely polymorphic:~\isa{{\isacharprime}a}.
|
wenzelm@12572
|
136 |
When a fixed record value is expressed using just its standard
|
wenzelm@12572
|
137 |
fields, the value of \isa{more} is implicitly set to \isa{{\isacharparenleft}{\isacharparenright}},
|
wenzelm@12572
|
138 |
the empty tuple, which has type \isa{unit}. Within the record
|
wenzelm@12572
|
139 |
brackets, you can refer to the \isa{more} field by writing \isa{{\isasymdots}} (three dots):%
|
wenzelm@12572
|
140 |
\end{isamarkuptext}%
|
wenzelm@12572
|
141 |
\isamarkuptrue%
|
wenzelm@12572
|
142 |
\isacommand{lemma}\ {\isachardoublequote}Xcoord\ {\isasymlparr}Xcoord\ {\isacharequal}\ a{\isacharcomma}\ Ycoord\ {\isacharequal}\ b{\isacharcomma}\ {\isasymdots}\ {\isacharequal}\ p{\isasymrparr}\ {\isacharequal}\ a{\isachardoublequote}\isanewline
|
wenzelm@12572
|
143 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
144 |
\isacommand{by}\ simp\isamarkupfalse%
|
wenzelm@12572
|
145 |
%
|
wenzelm@12572
|
146 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
147 |
This lemma applies to any record whose first two fields are \isa{Xcoord} and~\isa{Ycoord}. Note that \isa{{\isasymlparr}Xcoord\ {\isacharequal}\ a{\isacharcomma}\ Ycoord\ {\isacharequal}\ b{\isacharcomma}\ {\isasymdots}\ {\isacharequal}\ {\isacharparenleft}{\isacharparenright}{\isasymrparr}} is actually the same as \isa{{\isasymlparr}Xcoord\ {\isacharequal}\ a{\isacharcomma}\ Ycoord\ {\isacharequal}\ b{\isasymrparr}}.
|
wenzelm@12572
|
148 |
|
wenzelm@12572
|
149 |
The pseudo-field \isa{more} can be selected in the usual way, but
|
wenzelm@12572
|
150 |
the identifier must be qualified:%
|
wenzelm@12572
|
151 |
\end{isamarkuptext}%
|
wenzelm@12572
|
152 |
\isamarkuptrue%
|
wenzelm@12572
|
153 |
\isacommand{lemma}\ {\isachardoublequote}point{\isachardot}more\ cpt{\isadigit{1}}\ {\isacharequal}\ {\isasymlparr}col\ {\isacharequal}\ Green{\isasymrparr}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
154 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
155 |
\isacommand{by}\ {\isacharparenleft}simp\ add{\isacharcolon}\ cpt{\isadigit{1}}{\isacharunderscore}def{\isacharparenright}\isamarkupfalse%
|
wenzelm@12572
|
156 |
%
|
wenzelm@12572
|
157 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
158 |
We see that the colour attached to this \isa{point} is a
|
wenzelm@12572
|
159 |
(rudimentary) record in its own right, namely \isa{{\isasymlparr}col\ {\isacharequal}\ Green{\isasymrparr}}. In order to select or update \isa{col} in the above
|
wenzelm@12572
|
160 |
fragment, \isa{{\isasymlparr}col\ {\isacharequal}\ Green{\isasymrparr}} needs to be put back into the
|
wenzelm@12572
|
161 |
context of its parent type scheme, say as \isa{more} part of a
|
wenzelm@12572
|
162 |
\isa{point}.
|
wenzelm@12572
|
163 |
|
wenzelm@12572
|
164 |
To define generic operations, we need to know a bit more about
|
wenzelm@12572
|
165 |
records. Our declaration of \isa{point} above generated two type
|
wenzelm@12572
|
166 |
abbreviations:
|
wenzelm@12572
|
167 |
|
wenzelm@12572
|
168 |
\smallskip
|
wenzelm@12572
|
169 |
\begin{tabular}{l}
|
wenzelm@12572
|
170 |
\isa{point}~\isa{{\isacharequal}}~\isa{{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharcolon}\ int{\isacharcomma}\ Ycoord\ {\isacharcolon}{\isacharcolon}\ int{\isasymrparr}} \\
|
wenzelm@12572
|
171 |
\isa{{\isacharprime}a\ point{\isacharunderscore}scheme}~\isa{{\isacharequal}}~\isa{{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharcolon}\ int{\isacharcomma}\ Ycoord\ {\isacharcolon}{\isacharcolon}\ int{\isacharcomma}\ {\isasymdots}\ {\isacharcolon}{\isacharcolon}\ {\isacharprime}a{\isasymrparr}} \\
|
wenzelm@12572
|
172 |
\end{tabular}
|
wenzelm@12572
|
173 |
\smallskip
|
wenzelm@12572
|
174 |
|
wenzelm@12572
|
175 |
Type \isa{point} is for rigid records having exactly the two fields
|
wenzelm@12572
|
176 |
\isa{Xcoord} and~\isa{Ycoord}, while the polymorphic type \isa{{\isacharprime}a\ point{\isacharunderscore}scheme} comprises all possible extensions to those two
|
wenzelm@12572
|
177 |
fields, recall that \isa{unit\ point{\isacharunderscore}scheme} coincides with \isa{point}. For example, let us define two operations --- methods, if
|
wenzelm@12572
|
178 |
we regard records as objects --- to get and set any point's \isa{Xcoord} field.%
|
wenzelm@12572
|
179 |
\end{isamarkuptext}%
|
wenzelm@12572
|
180 |
\isamarkuptrue%
|
wenzelm@12572
|
181 |
\isacommand{constdefs}\isanewline
|
wenzelm@12572
|
182 |
\ \ getX\ {\isacharcolon}{\isacharcolon}\ {\isachardoublequote}{\isacharprime}a\ point{\isacharunderscore}scheme\ {\isasymRightarrow}\ int{\isachardoublequote}\isanewline
|
wenzelm@12572
|
183 |
\ \ {\isachardoublequote}getX\ r\ {\isasymequiv}\ Xcoord\ r{\isachardoublequote}\isanewline
|
wenzelm@12572
|
184 |
\ \ setX\ {\isacharcolon}{\isacharcolon}\ {\isachardoublequote}{\isacharprime}a\ point{\isacharunderscore}scheme\ {\isasymRightarrow}\ int\ {\isasymRightarrow}\ {\isacharprime}a\ point{\isacharunderscore}scheme{\isachardoublequote}\isanewline
|
wenzelm@12572
|
185 |
\ \ {\isachardoublequote}setX\ r\ a\ {\isasymequiv}\ r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isasymrparr}{\isachardoublequote}\isamarkupfalse%
|
wenzelm@12572
|
186 |
%
|
wenzelm@12572
|
187 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
188 |
Here is a generic method that modifies a point, incrementing its
|
wenzelm@12572
|
189 |
\isa{Xcoord} field. The \isa{Ycoord} and \isa{more} fields
|
wenzelm@12572
|
190 |
are copied across. It works for any record type scheme derived from
|
wenzelm@12572
|
191 |
\isa{point}, such as \isa{cpoint}:%
|
wenzelm@12572
|
192 |
\end{isamarkuptext}%
|
wenzelm@12572
|
193 |
\isamarkuptrue%
|
wenzelm@12572
|
194 |
\isacommand{constdefs}\isanewline
|
wenzelm@12572
|
195 |
\ \ incX\ {\isacharcolon}{\isacharcolon}\ {\isachardoublequote}{\isacharprime}a\ point{\isacharunderscore}scheme\ {\isasymRightarrow}\ {\isacharprime}a\ point{\isacharunderscore}scheme{\isachardoublequote}\isanewline
|
wenzelm@12572
|
196 |
\ \ {\isachardoublequote}incX\ r\ {\isasymequiv}\ {\isasymlparr}Xcoord\ {\isacharequal}\ Xcoord\ r\ {\isacharplus}\ {\isadigit{1}}{\isacharcomma}\isanewline
|
wenzelm@12572
|
197 |
\ \ \ \ Ycoord\ {\isacharequal}\ Ycoord\ r{\isacharcomma}\ {\isasymdots}\ {\isacharequal}\ point{\isachardot}more\ r{\isasymrparr}{\isachardoublequote}\isamarkupfalse%
|
wenzelm@12572
|
198 |
%
|
wenzelm@12572
|
199 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
200 |
Generic theorems can be proved about generic methods. This trivial
|
wenzelm@12572
|
201 |
lemma relates \isa{incX} to \isa{getX} and \isa{setX}:%
|
wenzelm@12572
|
202 |
\end{isamarkuptext}%
|
wenzelm@12572
|
203 |
\isamarkuptrue%
|
wenzelm@12572
|
204 |
\isacommand{lemma}\ {\isachardoublequote}incX\ r\ {\isacharequal}\ setX\ r\ {\isacharparenleft}getX\ r\ {\isacharplus}\ {\isadigit{1}}{\isacharparenright}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
205 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
206 |
\isacommand{by}\ {\isacharparenleft}simp\ add{\isacharcolon}\ getX{\isacharunderscore}def\ setX{\isacharunderscore}def\ incX{\isacharunderscore}def{\isacharparenright}\isamarkupfalse%
|
wenzelm@12572
|
207 |
%
|
wenzelm@12572
|
208 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
209 |
\begin{warn}
|
wenzelm@12572
|
210 |
If you use the symbolic record brackets \isa{{\isasymlparr}} and \isa{{\isasymrparr}},
|
wenzelm@12572
|
211 |
then you must also use the symbolic ellipsis, ``\isa{{\isasymdots}}'', rather
|
wenzelm@12572
|
212 |
than three consecutive periods, ``\isa{{\isachardot}{\isachardot}{\isachardot}}''. Mixing the ASCII
|
wenzelm@12572
|
213 |
and symbolic versions causes a syntax error. (The two versions are
|
wenzelm@12572
|
214 |
more distinct on screen than they are on paper.)
|
wenzelm@12572
|
215 |
\end{warn}%\index{records!extensible|)}%
|
wenzelm@12572
|
216 |
\end{isamarkuptext}%
|
wenzelm@12572
|
217 |
\isamarkuptrue%
|
wenzelm@12572
|
218 |
%
|
wenzelm@12572
|
219 |
\isamarkupsubsection{Record Equality%
|
wenzelm@12572
|
220 |
}
|
wenzelm@12572
|
221 |
\isamarkuptrue%
|
wenzelm@12572
|
222 |
%
|
wenzelm@12572
|
223 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
224 |
Two records are equal\index{equality!of records} if all pairs of
|
wenzelm@12572
|
225 |
corresponding fields are equal. Record equalities are simplified
|
wenzelm@12572
|
226 |
automatically:%
|
wenzelm@12572
|
227 |
\end{isamarkuptext}%
|
wenzelm@12572
|
228 |
\isamarkuptrue%
|
wenzelm@12572
|
229 |
\isacommand{lemma}\ {\isachardoublequote}{\isacharparenleft}{\isasymlparr}Xcoord\ {\isacharequal}\ a{\isacharcomma}\ Ycoord\ {\isacharequal}\ b{\isasymrparr}\ {\isacharequal}\ {\isasymlparr}Xcoord\ {\isacharequal}\ a{\isacharprime}{\isacharcomma}\ Ycoord\ {\isacharequal}\ b{\isacharprime}{\isasymrparr}{\isacharparenright}\ {\isacharequal}\isanewline
|
wenzelm@12572
|
230 |
\ \ \ \ {\isacharparenleft}a\ {\isacharequal}\ a{\isacharprime}\ {\isasymand}\ b\ {\isacharequal}\ b{\isacharprime}{\isacharparenright}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
231 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
232 |
\isacommand{by}\ simp\isamarkupfalse%
|
wenzelm@12572
|
233 |
%
|
wenzelm@12572
|
234 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
235 |
The following equality is similar, but generic, in that \isa{r}
|
wenzelm@12572
|
236 |
can be any instance of \isa{point{\isacharunderscore}scheme}:%
|
wenzelm@12572
|
237 |
\end{isamarkuptext}%
|
wenzelm@12572
|
238 |
\isamarkuptrue%
|
wenzelm@12572
|
239 |
\isacommand{lemma}\ {\isachardoublequote}r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isacharcomma}\ Ycoord\ {\isacharcolon}{\isacharequal}\ b{\isasymrparr}\ {\isacharequal}\ r{\isasymlparr}Ycoord\ {\isacharcolon}{\isacharequal}\ b{\isacharcomma}\ Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isasymrparr}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
240 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
241 |
\isacommand{by}\ simp\isamarkupfalse%
|
wenzelm@12572
|
242 |
%
|
wenzelm@12572
|
243 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
244 |
We see above the syntax for iterated updates. We could equivalently
|
wenzelm@12572
|
245 |
have written the left-hand side as \isa{r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isacharcomma}\ Ycoord\ {\isacharcolon}{\isacharequal}\ b{\isasymrparr}}.
|
wenzelm@12572
|
246 |
|
wenzelm@12572
|
247 |
Record equality is \emph{extensional}: \index{extensionality!for
|
wenzelm@12572
|
248 |
records} a record is determined entirely by the values of its
|
wenzelm@12572
|
249 |
fields.%
|
wenzelm@12572
|
250 |
\end{isamarkuptext}%
|
wenzelm@12572
|
251 |
\isamarkuptrue%
|
wenzelm@12572
|
252 |
\isacommand{lemma}\ {\isachardoublequote}r\ {\isacharequal}\ {\isasymlparr}Xcoord\ {\isacharequal}\ Xcoord\ r{\isacharcomma}\ Ycoord\ {\isacharequal}\ Ycoord\ r{\isasymrparr}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
253 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
254 |
\isacommand{by}\ simp\isamarkupfalse%
|
wenzelm@12572
|
255 |
%
|
wenzelm@12572
|
256 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
257 |
The generic version of this equality includes the field \isa{more}:%
|
wenzelm@12572
|
258 |
\end{isamarkuptext}%
|
wenzelm@12572
|
259 |
\isamarkuptrue%
|
wenzelm@12572
|
260 |
\isacommand{lemma}\ {\isachardoublequote}r\ {\isacharequal}\ {\isasymlparr}Xcoord\ {\isacharequal}\ Xcoord\ r{\isacharcomma}\ Ycoord\ {\isacharequal}\ Ycoord\ r{\isacharcomma}\ {\isasymdots}\ {\isacharequal}\ point{\isachardot}more\ r{\isasymrparr}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
261 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
262 |
\isacommand{by}\ simp\isamarkupfalse%
|
wenzelm@12572
|
263 |
%
|
wenzelm@12572
|
264 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
265 |
Note that the \isa{r} is of a different (more general) type than
|
wenzelm@12572
|
266 |
the previous one.
|
wenzelm@12572
|
267 |
|
wenzelm@12572
|
268 |
\medskip The simplifier can prove many record equalities
|
wenzelm@12572
|
269 |
automatically, but general equality reasoning can be tricky.
|
wenzelm@12572
|
270 |
Consider proving this obvious fact:%
|
wenzelm@12572
|
271 |
\end{isamarkuptext}%
|
wenzelm@12572
|
272 |
\isamarkuptrue%
|
wenzelm@12572
|
273 |
\isacommand{lemma}\ {\isachardoublequote}r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isasymrparr}\ {\isacharequal}\ r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isacharprime}{\isasymrparr}\ {\isasymLongrightarrow}\ a\ {\isacharequal}\ a{\isacharprime}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
274 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
275 |
\isacommand{apply}\ simp{\isacharquery}\isanewline
|
wenzelm@12572
|
276 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
277 |
\isacommand{oops}\isamarkupfalse%
|
wenzelm@12572
|
278 |
%
|
wenzelm@12572
|
279 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
280 |
The simplifier can do nothing, since general record equality is not
|
wenzelm@12572
|
281 |
eliminated automatically. One way to proceed is by an explicit
|
wenzelm@12572
|
282 |
forward step that applies the selector \isa{Xcoord} to both sides
|
wenzelm@12572
|
283 |
of the assumed record equality:%
|
wenzelm@12572
|
284 |
\end{isamarkuptext}%
|
wenzelm@12572
|
285 |
\isamarkuptrue%
|
wenzelm@12572
|
286 |
\isacommand{lemma}\ {\isachardoublequote}r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isasymrparr}\ {\isacharequal}\ r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isacharprime}{\isasymrparr}\ {\isasymLongrightarrow}\ a\ {\isacharequal}\ a{\isacharprime}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
287 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
288 |
\isacommand{apply}\ {\isacharparenleft}drule{\isacharunderscore}tac\ f\ {\isacharequal}\ Xcoord\ \isakeyword{in}\ arg{\isacharunderscore}cong{\isacharparenright}\isamarkupfalse%
|
wenzelm@12572
|
289 |
%
|
wenzelm@12572
|
290 |
\begin{isamarkuptxt}%
|
wenzelm@12572
|
291 |
\begin{isabelle}%
|
wenzelm@12572
|
292 |
\ {\isadigit{1}}{\isachardot}\ Xcoord\ {\isacharparenleft}r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isasymrparr}{\isacharparenright}\ {\isacharequal}\ Xcoord\ {\isacharparenleft}r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isacharprime}{\isasymrparr}{\isacharparenright}\ {\isasymLongrightarrow}\ a\ {\isacharequal}\ a{\isacharprime}%
|
wenzelm@12572
|
293 |
\end{isabelle}
|
wenzelm@12572
|
294 |
Now, \isa{simp} will reduce the assumption to the desired
|
wenzelm@12572
|
295 |
conclusion.%
|
wenzelm@12572
|
296 |
\end{isamarkuptxt}%
|
wenzelm@12572
|
297 |
\ \ \isamarkuptrue%
|
wenzelm@12572
|
298 |
\isacommand{apply}\ simp\isanewline
|
wenzelm@12572
|
299 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
300 |
\isacommand{done}\isamarkupfalse%
|
wenzelm@12572
|
301 |
%
|
wenzelm@12572
|
302 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
303 |
The \isa{cases} method is preferable to such a forward proof.
|
wenzelm@12572
|
304 |
State the desired lemma again:%
|
wenzelm@12572
|
305 |
\end{isamarkuptext}%
|
wenzelm@12572
|
306 |
\isamarkuptrue%
|
wenzelm@12572
|
307 |
\isacommand{lemma}\ {\isachardoublequote}r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isasymrparr}\ {\isacharequal}\ r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isacharprime}{\isasymrparr}\ {\isasymLongrightarrow}\ a\ {\isacharequal}\ a{\isacharprime}{\isachardoublequote}\isamarkupfalse%
|
wenzelm@12572
|
308 |
%
|
wenzelm@12572
|
309 |
\begin{isamarkuptxt}%
|
wenzelm@12572
|
310 |
The \methdx{cases} method adds an equality to replace the named
|
wenzelm@12572
|
311 |
record variable by an explicit record, listing all fields. It
|
wenzelm@12572
|
312 |
even includes the pseudo-field \isa{more}, since the record
|
wenzelm@12572
|
313 |
equality stated above is generic.%
|
wenzelm@12572
|
314 |
\end{isamarkuptxt}%
|
wenzelm@12572
|
315 |
\ \ \isamarkuptrue%
|
wenzelm@12572
|
316 |
\isacommand{apply}\ {\isacharparenleft}cases\ r{\isacharparenright}\isamarkupfalse%
|
wenzelm@12572
|
317 |
%
|
wenzelm@12572
|
318 |
\begin{isamarkuptxt}%
|
wenzelm@12572
|
319 |
\begin{isabelle}%
|
wenzelm@12572
|
320 |
\ {\isadigit{1}}{\isachardot}\ {\isasymAnd}Xcoord\ Ycoord\ more{\isachardot}\isanewline
|
wenzelm@12572
|
321 |
\isaindent{\ {\isadigit{1}}{\isachardot}\ \ \ \ }{\isasymlbrakk}r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isasymrparr}\ {\isacharequal}\ r{\isasymlparr}Xcoord\ {\isacharcolon}{\isacharequal}\ a{\isacharprime}{\isasymrparr}{\isacharsemicolon}\isanewline
|
wenzelm@12572
|
322 |
\isaindent{\ {\isadigit{1}}{\isachardot}\ \ \ \ \ \ \ }r\ {\isacharequal}\ {\isasymlparr}Xcoord\ {\isacharequal}\ Xcoord{\isacharcomma}\ Ycoord\ {\isacharequal}\ Ycoord{\isacharcomma}\ {\isasymdots}\ {\isacharequal}\ more{\isasymrparr}{\isasymrbrakk}\isanewline
|
wenzelm@12572
|
323 |
\isaindent{\ {\isadigit{1}}{\isachardot}\ \ \ \ }{\isasymLongrightarrow}\ a\ {\isacharequal}\ a{\isacharprime}%
|
wenzelm@12572
|
324 |
\end{isabelle}
|
wenzelm@12572
|
325 |
Again, \isa{simp} finishes the proof. Because \isa{r} has
|
wenzelm@12572
|
326 |
become an explicit record expression, the updates can be applied
|
wenzelm@12572
|
327 |
and the record equality can be replaced by equality of the
|
wenzelm@12572
|
328 |
corresponding fields (due to injectivity).%
|
wenzelm@12572
|
329 |
\end{isamarkuptxt}%
|
wenzelm@12572
|
330 |
\ \ \isamarkuptrue%
|
wenzelm@12572
|
331 |
\isacommand{apply}\ simp\isanewline
|
wenzelm@12572
|
332 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
333 |
\isacommand{done}\isamarkupfalse%
|
wenzelm@12572
|
334 |
%
|
wenzelm@12572
|
335 |
\isamarkupsubsection{Extending and Truncating Records%
|
wenzelm@12572
|
336 |
}
|
wenzelm@12572
|
337 |
\isamarkuptrue%
|
wenzelm@12572
|
338 |
%
|
wenzelm@12572
|
339 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
340 |
Each record declaration introduces functions to refer collectively
|
wenzelm@12572
|
341 |
to a record's fields and to convert between related record types.
|
wenzelm@12572
|
342 |
They can, for instance, convert between types \isa{point} and \isa{cpoint}. We can add a colour to a point or to convert a \isa{cpoint} to a \isa{point} by forgetting its colour.
|
wenzelm@12572
|
343 |
|
wenzelm@12572
|
344 |
\begin{itemize}
|
wenzelm@12572
|
345 |
|
wenzelm@12572
|
346 |
\item Function \cdx{make} takes as arguments all of the record's
|
wenzelm@12572
|
347 |
fields. It returns the corresponding record.
|
wenzelm@12572
|
348 |
|
wenzelm@12572
|
349 |
\item Function \cdx{fields} takes the record's new fields and
|
wenzelm@12572
|
350 |
returns a record fragment consisting of just those fields. This may
|
wenzelm@12572
|
351 |
be filled into the \isa{more} part of the parent record scheme.
|
wenzelm@12572
|
352 |
|
wenzelm@12572
|
353 |
\item Function \cdx{extend} takes two arguments: a record to be
|
wenzelm@12572
|
354 |
extended and a record containing the new fields.
|
wenzelm@12572
|
355 |
|
wenzelm@12572
|
356 |
\item Function \cdx{truncate} takes a record (possibly an extension
|
wenzelm@12572
|
357 |
of the original record type) and returns a fixed record, removing
|
wenzelm@12572
|
358 |
any additional fields.
|
wenzelm@12572
|
359 |
|
wenzelm@12572
|
360 |
\end{itemize}
|
wenzelm@12572
|
361 |
|
wenzelm@12572
|
362 |
These functions merely provide handsome abbreviations for standard
|
wenzelm@12572
|
363 |
record expressions involving constructors and selectors. The
|
wenzelm@12572
|
364 |
definitions, which are \emph{not} unfolded by default, are made
|
wenzelm@12572
|
365 |
available by the collective name of \isa{defs} (e.g.\ \isa{point{\isachardot}defs} or \isa{cpoint{\isachardot}defs}).
|
wenzelm@12572
|
366 |
|
wenzelm@12572
|
367 |
For example, here are the versions of those functions generated for
|
wenzelm@12572
|
368 |
record \isa{point}. We omit \isa{point{\isachardot}fields}, which happens to
|
wenzelm@12572
|
369 |
be the same as \isa{point{\isachardot}make}.
|
wenzelm@12572
|
370 |
|
wenzelm@12572
|
371 |
\begin{isabelle}%
|
wenzelm@12572
|
372 |
point{\isachardot}make\ {\isacharquery}Xcoord\ {\isacharquery}Ycoord\ {\isasymequiv}\ {\isasymlparr}Xcoord\ {\isacharequal}\ {\isacharquery}Xcoord{\isacharcomma}\ Ycoord\ {\isacharequal}\ {\isacharquery}Ycoord{\isasymrparr}\isanewline
|
wenzelm@12572
|
373 |
point{\isachardot}extend\ {\isacharquery}r\ {\isacharquery}more\ {\isasymequiv}\isanewline
|
wenzelm@12572
|
374 |
{\isasymlparr}Xcoord\ {\isacharequal}\ Xcoord\ {\isacharquery}r{\isacharcomma}\ Ycoord\ {\isacharequal}\ Ycoord\ {\isacharquery}r{\isacharcomma}\ {\isasymdots}\ {\isacharequal}\ {\isacharquery}more{\isasymrparr}\isanewline
|
wenzelm@12572
|
375 |
point{\isachardot}truncate\ {\isacharquery}r\ {\isasymequiv}\ {\isasymlparr}Xcoord\ {\isacharequal}\ Xcoord\ {\isacharquery}r{\isacharcomma}\ Ycoord\ {\isacharequal}\ Ycoord\ {\isacharquery}r{\isasymrparr}%
|
wenzelm@12572
|
376 |
\end{isabelle}
|
wenzelm@12572
|
377 |
|
wenzelm@12572
|
378 |
Contrast those with the corresponding functions for record \isa{cpoint}. Observe \isa{cpoint{\isachardot}fields} in particular.
|
wenzelm@12572
|
379 |
|
wenzelm@12572
|
380 |
\begin{isabelle}%
|
wenzelm@12572
|
381 |
cpoint{\isachardot}make\ {\isacharquery}Xcoord\ {\isacharquery}Ycoord\ {\isacharquery}col\ {\isasymequiv}\isanewline
|
wenzelm@12572
|
382 |
{\isasymlparr}Xcoord\ {\isacharequal}\ {\isacharquery}Xcoord{\isacharcomma}\ Ycoord\ {\isacharequal}\ {\isacharquery}Ycoord{\isacharcomma}\ col\ {\isacharequal}\ {\isacharquery}col{\isasymrparr}\isanewline
|
wenzelm@12572
|
383 |
cpoint{\isachardot}extend\ {\isacharquery}r\ {\isacharquery}more\ {\isasymequiv}\isanewline
|
wenzelm@12572
|
384 |
{\isasymlparr}Xcoord\ {\isacharequal}\ Xcoord\ {\isacharquery}r{\isacharcomma}\ Ycoord\ {\isacharequal}\ Ycoord\ {\isacharquery}r{\isacharcomma}\ col\ {\isacharequal}\ col\ {\isacharquery}r{\isacharcomma}\ {\isasymdots}\ {\isacharequal}\ {\isacharquery}more{\isasymrparr}\isanewline
|
wenzelm@12572
|
385 |
cpoint{\isachardot}truncate\ {\isacharquery}r\ {\isasymequiv}\isanewline
|
wenzelm@12572
|
386 |
{\isasymlparr}Xcoord\ {\isacharequal}\ Xcoord\ {\isacharquery}r{\isacharcomma}\ Ycoord\ {\isacharequal}\ Ycoord\ {\isacharquery}r{\isacharcomma}\ col\ {\isacharequal}\ col\ {\isacharquery}r{\isasymrparr}%
|
wenzelm@12572
|
387 |
\end{isabelle}
|
wenzelm@12572
|
388 |
|
wenzelm@12572
|
389 |
To demonstrate these functions, we declare a new coloured point by
|
wenzelm@12572
|
390 |
extending an ordinary point. Function \isa{point{\isachardot}extend} augments
|
wenzelm@12572
|
391 |
\isa{pt{\isadigit{1}}} with a colour, which is converted into an appropriate
|
wenzelm@12572
|
392 |
record fragment by \isa{cpoint{\isachardot}fields}.%
|
wenzelm@12572
|
393 |
\end{isamarkuptext}%
|
wenzelm@12572
|
394 |
\isamarkuptrue%
|
wenzelm@12572
|
395 |
\isacommand{constdefs}\isanewline
|
wenzelm@12572
|
396 |
\ \ cpt{\isadigit{2}}\ {\isacharcolon}{\isacharcolon}\ cpoint\isanewline
|
wenzelm@12572
|
397 |
\ \ {\isachardoublequote}cpt{\isadigit{2}}\ {\isasymequiv}\ point{\isachardot}extend\ pt{\isadigit{1}}\ {\isacharparenleft}cpoint{\isachardot}fields\ Green{\isacharparenright}{\isachardoublequote}\isamarkupfalse%
|
wenzelm@12572
|
398 |
%
|
wenzelm@12572
|
399 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
400 |
The coloured points \isa{cpt{\isadigit{1}}} and \isa{cpt{\isadigit{2}}} are equal. The
|
wenzelm@12572
|
401 |
proof is trivial, by unfolding all the definitions. We deliberately
|
wenzelm@12572
|
402 |
omit the definition of~\isa{pt{\isadigit{1}}} in order to reveal the underlying
|
wenzelm@12572
|
403 |
comparison on type \isa{point}.%
|
wenzelm@12572
|
404 |
\end{isamarkuptext}%
|
wenzelm@12572
|
405 |
\isamarkuptrue%
|
wenzelm@12572
|
406 |
\isacommand{lemma}\ {\isachardoublequote}cpt{\isadigit{1}}\ {\isacharequal}\ cpt{\isadigit{2}}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
407 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
408 |
\isacommand{apply}\ {\isacharparenleft}simp\ add{\isacharcolon}\ cpt{\isadigit{1}}{\isacharunderscore}def\ cpt{\isadigit{2}}{\isacharunderscore}def\ point{\isachardot}defs\ cpoint{\isachardot}defs{\isacharparenright}\isamarkupfalse%
|
wenzelm@12572
|
409 |
%
|
wenzelm@12572
|
410 |
\begin{isamarkuptxt}%
|
wenzelm@12572
|
411 |
\begin{isabelle}%
|
wenzelm@12572
|
412 |
\ {\isadigit{1}}{\isachardot}\ Xcoord\ pt{\isadigit{1}}\ {\isacharequal}\ {\isadigit{9}}{\isadigit{9}}{\isadigit{9}}\ {\isasymand}\ Ycoord\ pt{\isadigit{1}}\ {\isacharequal}\ {\isadigit{2}}{\isadigit{3}}%
|
wenzelm@12572
|
413 |
\end{isabelle}%
|
wenzelm@12572
|
414 |
\end{isamarkuptxt}%
|
wenzelm@12572
|
415 |
\ \ \isamarkuptrue%
|
wenzelm@12572
|
416 |
\isacommand{apply}\ {\isacharparenleft}simp\ add{\isacharcolon}\ pt{\isadigit{1}}{\isacharunderscore}def{\isacharparenright}\isanewline
|
wenzelm@12572
|
417 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
418 |
\isacommand{done}\isamarkupfalse%
|
wenzelm@12572
|
419 |
%
|
wenzelm@12572
|
420 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
421 |
In the example below, a coloured point is truncated to leave a
|
wenzelm@12572
|
422 |
point. We must use the \isa{truncate} function of the shorter
|
wenzelm@12572
|
423 |
record.%
|
wenzelm@12572
|
424 |
\end{isamarkuptext}%
|
wenzelm@12572
|
425 |
\isamarkuptrue%
|
wenzelm@12572
|
426 |
\isacommand{lemma}\ {\isachardoublequote}point{\isachardot}truncate\ cpt{\isadigit{2}}\ {\isacharequal}\ pt{\isadigit{1}}{\isachardoublequote}\isanewline
|
wenzelm@12572
|
427 |
\ \ \isamarkupfalse%
|
wenzelm@12572
|
428 |
\isacommand{by}\ {\isacharparenleft}simp\ add{\isacharcolon}\ pt{\isadigit{1}}{\isacharunderscore}def\ cpt{\isadigit{2}}{\isacharunderscore}def\ point{\isachardot}defs{\isacharparenright}\isamarkupfalse%
|
wenzelm@12572
|
429 |
%
|
wenzelm@12572
|
430 |
\begin{isamarkuptext}%
|
wenzelm@12572
|
431 |
\begin{exercise}
|
wenzelm@12572
|
432 |
Extend record \isa{cpoint} to have a further field, \isa{intensity}, of type~\isa{nat}. Experiment with coercions among the
|
wenzelm@12572
|
433 |
three record types.
|
wenzelm@12572
|
434 |
\end{exercise}
|
wenzelm@12572
|
435 |
|
wenzelm@12572
|
436 |
\begin{exercise}
|
wenzelm@12572
|
437 |
(For Java programmers.)
|
wenzelm@12572
|
438 |
Model a small class hierarchy using records.
|
wenzelm@12572
|
439 |
\end{exercise}
|
wenzelm@12572
|
440 |
\index{records|)}%
|
wenzelm@12572
|
441 |
\end{isamarkuptext}%
|
wenzelm@12572
|
442 |
\isamarkuptrue%
|
wenzelm@12572
|
443 |
\isamarkupfalse%
|
wenzelm@12572
|
444 |
\end{isabellebody}%
|
wenzelm@12572
|
445 |
%%% Local Variables:
|
wenzelm@12572
|
446 |
%%% mode: latex
|
wenzelm@12572
|
447 |
%%% TeX-master: "root"
|
wenzelm@12572
|
448 |
%%% End:
|