1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
|
<html>
<head><title>NOTE-2-6-GUARDS.html -- ACL2 Version 3.1</title></head>
<body text=#000000 bgcolor="#FFFFFF">
<h3>NOTE-2-6-GUARDS</h3>ACL2 Version 2.6 Notes on Guard-related Changes
<pre>Major Section: <a href="NOTE-2-6.html">NOTE-2-6</a>
</pre><p>
When you <code><a href="DECLARE.html">declare</a></code> that a function treats certain formals
as <code>:</code><code><a href="STOBJ.html">stobj</a></code><code>s</code>, the <a href="GUARD.html">guard</a> of the function is automatically
extended to include the corresponding stobj-recognizer calls. For example,
if a definition includes <code>(declare (xargs :stobjs (ST)))</code> then the
guard of the function is changed by the addition of the conjunct
<code>(ST-P ST)</code>.<p>
One impact of this is that if you use the built-in ACL2 <code><a href="STATE.html">state</a></code>
as a formal parameter of a function, <code>(STATE-P STATE)</code> is added to
the guard. This may introduce a guard where there was none in
previous versions of the system. In older versions, therefore, no
attempt would be made to <code><a href="VERIFY-GUARDS.html">verify-guards</a></code>, while in the new
version, we would attempt guard verification. You may wish to add
<code>(declare (xargs :verify-guards nil))</code> to such definitions.<p>
A related change affects users who do not use stobjs or <code>state</code>.
In previous versions of the system -- as now -- a <code>type</code>
declaration extended the guard you provided explicitly. Thus, if
you wrote <code>(declare (type integer n))</code> then <code>(INTEGERP n)</code> was
added to your guard. This is still the case and <code>:stobjs</code>
recognizers are similarly added. But in older versions of the system
we ``added'' the conjuncts without checking whether they were already
present in the guard you provided. This sometimes produced such
guards as <code>(and (integerp n) (integerp n))</code> where the first was
produced by your <code>type</code> declaration and the second was your
<code>:guard</code>. We now eliminate redundant conjuncts; this may rearrange
the order of the conjuncts.<p>
The guard conjectures for functions using <code>stobj</code>s have been simplified
somewhat by taking advantage of the syntactic restrictions checked for
single-threaded objects. <p>
The following functions have been modified so that character and
string arguments are restricted to standard characters.
(See <a href="STANDARD-CHAR-P.html">standard-char-p</a> and see <a href="STANDARD-CHAR-LISTP.html">standard-char-listp</a>.)
<blockquote><p>
<code><a href="UPPER-CASE-P.html">upper-case-p</a></code>
<code><a href="LOWER-CASE-P.html">lower-case-p</a></code>
<code><a href="CHAR-UPCASE.html">char-upcase</a></code>
<code><a href="CHAR-DOWNCASE.html">char-downcase</a></code>
<code>string-downcase1</code>
<code><a href="STRING-DOWNCASE.html">string-downcase</a></code>
<code>string-upcase1</code>
<code><a href="STRING-UPCASE.html">string-upcase</a></code>
<code><a href="CHAR-EQUAL.html">char-equal</a></code>
<code>string-equal1</code>
<code><a href="STRING-EQUAL.html">string-equal</a></code><p>
</blockquote>
Also, function <code><a href="STANDARD-STRING-ALISTP.html">standard-string-alistp</a></code> replaces function
<code>string-alistp</code>, with concomitant changes in the guard to
<code><a href="ASSOC-STRING-EQUAL.html">assoc-string-equal</a></code>, and in variable <code>*acl2-exports*</code>.
Also, lemma <code>standard-string-alistp-forward-to-alistp</code> replaces
lemma <code>string-alistp-forward-to-alistp</code>. There is a new lemma
<code>standard-char-p-nth</code>, which has also been added to <code>*acl2-exports*</code>.<p>
The guard had been inadvertently omitted from the definition of the
function <code><a href="SUBSTITUTE.html">substitute</a></code> (and its subroutine <code>substitute-ac</code>). This
omission has been corrected; also, the guard is slightly stronger
than the documentation had claimed (and that has been corrected).<p>
<p>
<br><br><br><a href="acl2-doc.html"><img src="llogo.gif"></a> <a href="acl2-doc-index.html"><img src="index.gif"></a>
</body>
</html>
|
