On 31/12/2010, at 12:02 AM, Dierk König <[email protected]> wrote:
>>> we shouldn't call this 'testing'.
>>>
>>> If anything, it is 'excercising' the application.
>>
>> Testing, or exercising, the application is such a common activity at each
>> stage of the application lifecycle. As this thread makes clear, it's about
>> much more than just running some unit tests when you build your jar.
>
> Testing is part of the build. That goes without saying.
>
> But calling the application with various inputs (here: timeouts) until it
> occasionally responds correctly once (!) doesn't deserve any special place in
> the build.
> The next time, you call the same functionality with the same parameters, it
> may just fail again
> because the server is under higher load or now serves _two_ users or so...
>
> The maximum info that you can get out of this approach is: "works
> occasionally but you cannot rely on it".
>
> I know, people do it, but that doesn't mean that it is the right thing to do
> and that we should contribute to misleading users into thinking there would
> be "tests" in the build.
Your argument seems predominantly theoretical and textbook based.
Gradle prides itself on being a tool with comprehensive support for solving
common problems. I fail to see how wanting to test web applications by
asserting browser behaviour is not a common problem.
---------------------------------------------------------------------
To unsubscribe from this list, please visit:
http://xircles.codehaus.org/manage_email
