???
Client configuration identification
(OS, browser, language, plug-ins, cookies, screen resolution,
color depth). You may find it advisable to identify a limited number of sets, only one of which
will be tested exhaustively. Once this is shown to be working the others can be tested. See
www.codehouse.com/browser_watch/ for some statistics on OS, Browser, plug-ins, cookies, screen
resolution, and color depth. Remember to start with something small which can unequivocally
be shown to work (or not) and which can then be scaled up.
???
Test coverage determination
. Identify some basis on which you can report test coverage. See
section 5.5.11 below.
???
Test ordering
. This is strategy-dependent (see Table 5.1) and depends on the degree your website
is dominated by web applications. Ensure that (for example) you answer one performance question
conclusively and unquestionably before attempting to answer another.
???
Test environment determination
. Ensure that your environment doesn??™t either affect other users
or can be affected by them. You need to be completely separate from developers and the real world
to be sure that it hasn??™t been polluted.
???
Data source determination
. Ensure the data is as real or valid as possible. In the worst case (the
site is totally new, the user profile ranges from nuns to terrorists, the numbers as vague as a
politician??™s promise) look for the worst case (the entire population of Brazil attempts to download
the HD version of ???The Sound of Music,??? in Chinese at 0500 GMT on Sunday (say)), and
70
Manage Software Testing
extrapolate from that.
Pages:
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207