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
|
Suggestions and tips for writing test files
Barton Willis
(1) test file contains pairs of expressions. The first
expression is the input, and the second is the expected output; for
example, here are two tests:
x + 0;
x$
x - x;
0$
You may terminate each line with either a semicolon or a dollar
sign. To keep the file somewhat organized, terminate the input line
with a semicolon and the expected output with a dollar sign. It's helpful to
separate each input / expected output pair with a blank line.
(2) Name your file "rtest_<a word that describes its contents>.mac".
For example, "rtest_abs.mac" is a good name for tests for the absolute
value function.
(3) If the input is something that doesn't need to be checked, make a
compound statement with the last statement being 0. Thus
(assume(x > 0), 0);
0$
abs(x);
x$
(4) If a test needs to use a non-default value for an option variable, try to
make the change local instead of global. For example
is (x < 1), prederror : false;
unknown$
Doing so keeps your test from modifying the global environment.
(5) The first line of the test file should be
(kill(all),0);
0$
(6) From a Maxima input prompt, use the 'batch' function to run your test:
(%i1) batch("rtest_abs.mac", 'test);
Error log on #<output stream rtest_abs.ERR>
You may need to give Maxima a full pathname to the file. After the
test finishes, you can look at the .ERR file to see which tests
failed.
(7) If your test defines functions, makes assumptions, or assigns
values to symbols, the last few lines of your test should clean up the
global environment; thus if your test does assume(x > 0), assigns a
value to 'a', and defines a function 'f', the last line of your test
should be something like
(forget(x > 0), remvalue(a), remfunction(f),0);
0$
(8) To append a test file to the main Maxima test suite,
you'll need to append the file name to the list in the file
/tests/testsuite.lisp. If your test has known errors, you'll need to
include this data into testsuite.lisp. To illustrate, the tests
42 and 42 in rtest_abs are known bugs. To tell Maxima that these
are known bugs, append ((mlist) "rtest_abs" 42 43) to the file
list in testsuite.lisp.
Also, you'll need to append the name of the test file to
makefile.am. Finally, build Maxima and run the test suite. If all goes
well, commit the new test file.
(9) Other:
(a) Check that your test runs multiple times without error. After
appending a new test to Maxima's test suite, make sure that run_testsuite()
runs multiple times without error.
(b) Always test the simple cases: abs(0), abs(0.0), abs(0.0b0), ... Also,
check that functions work correctly for CRE expressions, arguments that
contain '%i', empty lists, empty matrices, ... Thus always test
the 'boundary' cases; these are things like max(), min(), apply("*",[]), ....
(c) Check the sourceforge bug list for all reported bugs for the functions(s)
you are testing. Include tests for these bugs; also put a comment in your
test file that references the bug report:
/* See SF Bug # 771061 */
expand((vt . a^^(-1) . u+1)^^(-2));
((vt.a^^(-1).u)^^2+2*(vt.a^^(-1).u)+1)^^(-1)$
If a bug causes a fatal error, you'll need to exclude it from your test
(or better, include it, but comment it out).
(d) It might be easier to place all known failures at the top of the
file. With this policy, you can append new tests without updating the
list of known failures in testsuite.lisp.
(e) If a test checks a not so well known identity, include a reference to the
identity:
/* A&S 13.6.1 */
hgfred([v+1/2],[2*v+1],2*%i*z);
4^(v/2)*bessel_j(v,z)*gamma(v+1)*exp(%i*z)/z^v$
(f) Maxima evaluates the input, but only simplifies the expected output. Thus
a test such as
makelist(sin(k * %pi),k,1,5);
makelist(0,k,1,5)$
will fail. You'll need to write the test as either
makelist(sin(k * %pi),k,1,5);
''(makelist(0,k,1,5))$
or
makelist(sin(k * %pi),k,1,5);
[0,0,0,0,0]$
To get a test to pass,you may need to insert a few parenthesis in the
expected output; for example
-a/2;
-(a/2)$
Another way to handle such things is to use explicit calls to ratsimp:
ratsimp(-a/2);
''(ratsimp(-a/2))$
To test a function such as factor, you do not want to apply ratsimp.
|
