> I do apologise if I came across flippant or dismissive.
accepted ;-)
> I merely don't understand how the ideas you put forward manifest in terms of
> Gradle's implementation of verifying your software works.
My suggestion is to differentiate between verifications and the rest (which I
called "exercising" for the lack of a better word).
Whatever we can verify reliably is part of the testing and its results go into
the test-results. Every error breaks the build.
The rest should Gradle-vise be called from an extra task, lets say
gradle exercise
Results should be reported separately and errors may not break the build.
Note the on a CI or elsewhere , you are still free to start your build via
gradle build exercise
The additional reports are of informative character, just as CodeNarc or
coverage reports.
Beside the methodological arguments, there is a very mundane reason for this
distinction:
it avoids "false negatives" to creep into the tests, which then undermine the
whole testing effort.
"Yes, the tests failed, but that's ok..." is a phrase that I have heard far too
often and has invariably been
the beginning of actual test errors being dismissed and the final delivery of
unreliable applications.
happy new year everybody
Dierk
---------------------------------------------------------------------
To unsubscribe from this list, please visit:
http://xircles.codehaus.org/manage_email
