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
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
|
<html>
<head>
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
<meta http-equiv="Content-Language" content="en" />
<title>execline: the eltest program</title>
<meta name="Description" content="execline: the eltest program" />
<meta name="Keywords" content="execline command eltest test" />
<!-- <link rel="stylesheet" type="text/css" href="//skarnet.org/default.css" /> -->
</head>
<body>
<p>
<a href="index.html">execline</a><br />
<a href="//skarnet.org/software/">Software</a><br />
<a href="//skarnet.org/">skarnet.org</a>
</p>
<h1> The <tt>eltest</tt> program </h1>
<p>
eltest evaluates an expression and indicates the result via its
exit status.
</p>
<h2> Interface </h2>
<pre>
eltest <em>expression...</em>
</pre>
<p>
<tt>eltest</tt> acts as the generic POSIX
<a href="https://pubs.opengroup.org/onlinepubs/9699919799/utilities/test.html">test</a> utility,
but it diverges from the specification on how it parses ambiguous arguments, see below.
</p>
<p>
<tt>eltest</tt> supports all the standard
<a href="https://pubs.opengroup.org/onlinepubs/9699919799/utilities/test.html">test</a>
operands, plus all the extensions from
<a href="https://man7.org/linux/man-pages/man1/test.1.html">GNU test</a>, plus a few
extensions from the <tt>test</tt> builtin from
<a href="https://www.gnu.org/savannah-checkouts/gnu/bash/manual/bash.html#Bash-Conditional-Expressions">bash</a>.
The extensions to POSIX <tt>test</tt> are listed below.
</p>
<p>
<tt>eltest</tt> accepts an arbitrary number of arguments and, if the expression is
valid, always returns the result of the expression no matter how complex it
is.
</p>
<h2> Exit codes </h2>
<ul>
<li> 0: the test is true </li>
<li> 1: the test is false </li>
<li> 100: wrong usage </li>
<li> 101: internal error (should never happen, warrants a bug-report) </li>
<li> 111: system call failure </li>
</ul>
<h2> Posixness </h2>
<p>
<tt>eltest</tt> <strong>is not</strong> suitable as a Single Unix
<a href="https://pubs.opengroup.org/onlinepubs/9699919799/utilities/test.html">test</a>
program, due to the way it disambiguates between arguments and operators, see below.
However, if you never use arguments that start with a backslash, or that have the
same name as an existing operator, then
<tt>eltest</tt> exhibits the same behaviour as <tt>test</tt>.
</p>
<h2> Extensions to POSIX </h2>
<ul>
<li> <tt><em>expr1</em> -a <em>expr2</em></tt> :
tests whether <em>expr1</em> <strong>and</strong> <em>expr2</em> are true.
If <em>expr1</em> is false, then <em>expr2</em> is not evaluated. </li>
<li> <tt><em>expr1</em> -o <em>expr2</em></tt> :
tests whether <em>expr1</em> <strong>or</strong> <em>expr2</em> is true.
If <em>expr1</em> is true, then <em>expr2</em> is not evaluated. </li>
<li> <tt>-k <em>file</em></tt> : tests whether <em>file</em>
has the sticky bit. </li>
<li> <tt>-O <em>file</em></tt> : tests whether <em>file</em>
is owned by the effective uid of the current process. </li>
<li> <tt>-U <em>file</em></tt> : same. </li>
<li> <tt>-G <em>file</em></tt> : tests whether <em>file</em>'s gid
is the effective gid of the current process. </li>
<li> <tt>-N <em>file</em></tt> : tests whether <em>file</em> exists
and has been modified since it was last read. </li>
<li> <tt><em>file1</em> -nt <em>file2</em></tt> :
tests whether <em>file1</em> has a (strictly) newer modification date than <em>file2</em>. </li>
<li> <tt><em>file1</em> -ot <em>file2</em></tt> :
tests whether <em>file1</em> has a (strictly) older modification date than <em>file2</em>. </li>
<li> <tt><em>file1</em> -ef <em>file2</em></tt> :
tests whether <em>file1</em> and <em>file2</em> are physically the same
file (same device and inode numbers). </li>
<li> <tt>-v <em>var</em></tt> : tests whether the
<em>var</em> variable is defined in the current environment. </li>
<li> <tt><em>string</em> =~ <em>pattern</em></tt> :
tries to match <em>string</em> against extended regular expression
<em>pattern</em>. True if any part of <em>string</em> matches <em>pattern</em>;
in order to match whole strings, you must anchor <em>pattern</em> with
<tt>^</tt> and <tt>$</tt> markers. </li>
</ul>
<h2> Argument disambiguation </h2>
<p>
Unlike <a href="https://pubs.opengroup.org/onlinepubs/9699919799/utilities/test.html">test</a>,
which has different fixed syntax trees depending on the number of arguments it receives and
has undefined behaviour when called with more than 5 arguments, <tt>eltest</tt> accepts any
number of arguments and builds its syntax trees on the fly. This means that expressions such
as <tt>-n = -n</tt> cannot be automatically disambiguated: <tt>eltest</tt> does not know that
there are 3 arguments, so when it reads the first <tt>-n</tt> it assumes that it is an unary
operator, then when it reads <tt>=</tt> it assumes it is the argument to <tt>-n</tt>, then
when it reads the second <tt>-n</tt> it exits with a syntax error.
</p>
<p>
Doing otherwise would result in a combinatory explosion of possible syntax trees, making
it easy for users to trigger unbounded RAM consumption, and turning a simple utility into
a programming nightmare. This is why POSIX
<a href="https://pubs.opengroup.org/onlinepubs/9699919799/utilities/test.html">test</a>
is so restricted. But we don't want the same restrictions.
</p>
<p>
So, instead, <tt>eltest</tt> provides the user with a mechanism to make sure that
operands are never mistaken for operators:
</p>
<ul>
<li> An word that looks like an operator will always be interpreted like an operator.
So, expressions like <tt>-n = -n</tt> will result in a syntax error, because the
first <tt>-n</tt> will never be understood as data for the <tt>=</tt> operator. </li>
<li> A word that starts with a <tt>\</tt> (backslash) will always be interpreted
like data, never like an operator, and the backslash will be removed. This
means: <tt>\-n = \-n</tt> is a valid expression testing the equality between
the strings <tt>-n</tt> and </tt>-n</tt>.
<ul>
<li> Be aware that execline as well as the shell use one backlash for their own
unquoting mechanism, so when using backslashes in an execline or shell script, they
must be doubled. You would probably need to type something like <tt>\\-n = \\-n</tt>.
</ul> </li>
<li> So, if your script tests equality between <tt>$a</tt> and <tt>$b</tt>, and there's
a possiblity that the contents of these variables look like <tt>eltest</tt> operators,
the proper syntax would be: <tt>eltest \\${a} = \\${b}</tt>. </li>
</ul>
<p>
Note that these details are irrelevant to a huge majority of <tt>eltest</tt> use
cases, because most of the time users only need a simple test
such as <tt>eltest -r ${file}</tt> to check that <tt>$file</tt> is readable, and
there's no possible ambiguity. So don't panic over this.
</p>
<h2> Notes </h2>
<ul>
<li> <tt>eltest</tt> is a replacement for the ill-named, and now deprecated,
<a href="https://skarnet.org/software/s6-portable-utils/s6-test.html">s6-test</a>
program, part of the (just as ill-named)
<a href="https://skarnet.org/software/s6-portable-utils/">s6-portable-utils</a>
package. It is too valuable a utility to be part of a marginal package, and
has nothing to do with <a href="https://skarnet.org/software/s6/">s6</a>. </li>
</ul>
</body>
</html>
|
