Frode Nordahl wrote:
On 25. sep. 2006, at 10.38, Ian Eiloart wrote:
>
Alternatively, the test suite (is there one?) should check that a
message from each time zone can be read. You won't get a tester in
every time zone.
No test suite can ever cover all those corner cases that a program will
be subject to. They will probably help weeding out the relatively
obvious problems, but "new" problems like this one will not be avoided
the first time they occur. Having a test for it in place will of course
avoid the same bug appearing in a future release.
My motto: "Never have the same bug twice." Regression testing is, in my
opinion, more valuable than /random tests/ of every feature deemed
likely to break.
I think it would really help to announce and release a beta or release
candidate some time before the release, as this most certainly will make
more people test and hopefully find more of these problems before the
final release hits the press.
I'd like to second this motion. Calling for extensive testing of a
"release candidate" is common practice in some other Internet-based
software projects, and perhaps this strategy would be successful here as
well.
_______________________________________________
Imap-uw mailing list
[email protected]
https://mailman1.u.washington.edu/mailman/listinfo/imap-uw
