README.txt
1 iExploder 1.3.2
2 ===============
3
4 Welcome to iExploder. a highly inefficient, but fairly effective web
5 browser tester. The code still has a lot of work to be done, but it's
6 definitely usable. Here are some notable features:
7
8 * Tests all common HTML and CSS tags and attributes, as parsed from
9 the KHTML, WebKit and Mozilla source trees, as well as tags for
10 Internet Explorer from MSDN. This also includes a few Javascript hooks.
11 * Numeric, and String overflow and formatting tests
12 * Sequential and Randomized Test Case Generation
13 * Test Case Lookups
14 * Subtest generation
15
16
17 Installation (Standalone)
18 -------------------------
19 Make sure you have Ruby installed (comes with Mac OS X, most Linux
20 distributions). See http://www.ruby-lang.org/ if you do not.
21
22 If you do not already have a webserver setup, you can use the server
23 built into iexploder. Simply go into the htdocs directory and type:
24
25 % ruby webserver.rb
26
27 A webserver will then start on port 2000 with the iexploder form. If
28 port 2000 is not preferable, you can pass it another port on the command
29 line:
30
31 % ruby webserver.rb 2001
32
33 Please note that lasthit.rb does not currently work with the logs output
34 from this method. I recommend using a seperate instance/port number
35 for each browser you test simultaneous using this method.
36
37
38 Installation (External Webserver)
39 ---------------------------------
40 If you wish to use an external webserver (required for lasthit.rb use),
41 you may do so. IExploder has been tested with apache.
42
43 Copy the contents of the htdocs/ folder to any directory served
44 by your webserver. Make sure that directory can execute CGI scripts. If
45 performance seems to be low, please try using mod_ruby.
46
47
48 FAQ:
49 ----
50 1) Are the tests always the same?
51
52 The test cases should always be the same on a single installation, but not
53 necessarily on different installations of iExploder. Random generator seeds
54 may differ between operating systems and platforms. If you alter the tag and
55 property counts in config.rb, it will change the test cases as well.
56
57
58 2) How do I look up the last successful test for a client?
59
60 Use tools/lasthit.rb. When I get a crash, I usually do something like:
61
62 % tail -15000 /var/log/apache2/access_log | ./lasthit.rb
63
64 Letting you know how many tests and what the last test id was for each
65 client tested. You can then try to repeat the test, or go through the
66 subtests to see if you can repeat the crash.
67
68
69 3) How do subtests work?
70
71 If you see a crash on a particular test, and would like to determine the exact
72 line that is crashing it, you can use subtests. To do so, go back to the test
73 submission form, and start the test from the number that a crash was indicated
74 on. Instead of leaving the "subtest" field blank, set it to 1. This will rotate
75 through each subtest for a particular test.
76
77 Each subtest will rotate through a tag offset and a number of tags to
78 garble, which should help you isolate the instance. The number of tags
79 used doubles each cycle. Here is an idea of how many subtests to expect
80 based on your $HTML_MAX_TAGS settings:
81
82 tags subtests
83 ----------------
84 32 138
85 48 236
86 64 332
87 96 558
88 128 782
89
90 Most of the time you will be able to replicate a crash within the first
91 $HTML_MAX_TAGS subtests, but sometimes crashes are due to a combination
92 of corrupted tags.
93
94
95 4) How come I can't seem to repeat the crash?
96
97 Many browser crashes are race conditions that are not easy to repeat. Some
98 crashes only happen when going from test 4 -> test 5 -> test 6. If you can't
99 repeat the crash through subtests or a lookup of the failing test, try going
100 back a few tests.
101
102 That said, some crashes are due to race conditions that are very difficult
103 to replicate.
104
105
106 5) Why did you write this?
107
108 I wanted to make sure that FireFox had as many bugs fixed in it as possible
109 before the 1.0 release. After 1.0 came out, I kept improving it.
110
111
112 6) Why does Internet Explorer run the tests so slowly?
113
114 <META> refresh tags are very fragile in Internet Explorer, and can be easily
115 be rendered useless by other tags on the page. If this happens, a javascript
116 refresh will execute after a 1 second delay.
117
118
119
120 7) How do I change the number of tags iExploder tests per page?
121
122 See config.rb. I personally recommend 32-128 HTML tags per page. While this
123 seems to be a lot to go through when designing a test case, that's why the
124 subtest engine was made. Different web browsers will have different
125 performance characteristics when it comes to the number of tags per page.
126
127 Here are the results with Firefox 2.0b1 (Bon Echo) and the iExploder
128 built-in webserver running tests 1-250.
129
130 tags seconds pages/second tags/second
131 -----------------------------------------
132 32 60 4.0 131
133 48 85 2.9 141
134 64 95 2.6 168
135 96 120 2.1 200 *DEFAULT*
136 128 140 1.8 228
137 196 228 1.1 210
138 256 308 0.8 207
139
140 If you find pages/second to be more important than tags/second, I would
141 change $HTML_MAX_TAGS to 32. Do keep in mind that large tag counts mean
142 longer subtest generation periods.
143
144
145 8) What other performance enhancements can I make?
146
147 * Before using iExploder, reset your browser history
148 * Minimize your browser while iExploder is running
149 * If using Apache, make use of mod_ruby
150