Sebastien Roy wrote: > On Fri, 2009-06-12 at 13:39 -0700, Valerie Bubb Fenwick wrote: >> Sorry for the delay in following up with you. > > And I'm reciprocating the delay. :-) > >> As you can now see, >> the RTI nits page has been updated with the language agreed upon >> with the ON test sponsor folks, which happened before your mail came >> out. >> >> I believe your second bullet is adequately covered in the new >> nits text, but I don't think the first is fully covered. Could you >> look at the new text at let me know what, if any, additional testing >> text you want? > > My initial concern had to do with a flurry of RTIs that had no testing > details in them (i.e. "I did some testing and it passed"). In most > cases the RE has done proper testing, but in any case there's simply no > way for the RTI advocate to evaluate the extent of the testing and make > a decision regarding its adequacy. There needs to be documentation of > what was tested and how, and the results of the testing. It should be > shrink-to-fit, but the basic information should be there. > > I'd suggest something like: > > * Test Results > * This section should contain enough information for the CRT > advocate to evaluate if testing was adequate. Pointers to logs of > automated tests or terminal output of manual tests are welcome, > along with a high-level summary. For a simple fix, a description > of the testing methodology is fine. In all cases, a statement of > the form, "I tested the code and it passed" is inadequate.
That's more explicit than the existing text: "Developers are responsible for validating their changes prior to integration. This work should be described in the 'Test Results' section of the RTI." Will it drive the desired behavior? Hard to tell. I'm fine with the suggested change. It seems like "rti nits" is evolving into "how to fill out an RTI conscientiously." :) --Mark
