Michael G Schwern wrote: > On 8/15/06, Adrian Howard <[EMAIL PROTECTED]> wrote: >> To be honest, I'm not entirely sure that I could come up with a >> really convincing argument for TAP for an audience using JUnit, >> TestNG and similar... >> >> Do we have a convincing argument? Maybe I'm just being overly >> pessimistic today :-) > > The separation of the display code from the tests themselves is a > bonus. Though one would have to check with folks who actually use > XUnit to be sure this is a problem, I can imagine cases where the test > author used some crappy XUnit implementation and the display of the > tests sucks (its ambiguous, missing information, doesn't provide > enough information, only graphical, etc). And there's little you can > do about that since the test and the display code are bound.
I think this is the most important benefit. Test runs aren't bound to the single time it was run with a certain harness. They can be captured and viewed later by the same developer or shared among developers. If the report is descriptive enough (catching STDERR, etc) then it can be a great tool for debugging and would go hand-in-hand with a bug tracker. Imagine getting a bug report of your test suite failing along with the full TAP output of the test run. [snip] > There's also the advantage of it being cross-platform for those > projects which involve multiple languages (Java and PHP maybe?) > > Finally, XUnit can be coerced into outputting TAP. > > Trouble is at the moment all this is still in the prototype stage. > And none of them are killer. I'm actually hopeful that Smolder (http://sourceforge.net/projects/smolder) or something like it will show how useful abstracting out the test output from the harness can be. I'm planning on doing a 1.0 release of it pretty soon if anyone's interested. -- Michael Peters Developer Plus Three, LP
