On 11/26/2013 08:10 PM, Josh Matthews wrote:
* regressionwindow-wanted, testcase-wanted: regression from earlier versions is present, and some help crafting a testcase/clear steps to consistently reproduce is desired
I think we should keep "testcase" and "steps to reproduce" as separate notions. There is a keyword to flag the absense of the latter: steps-wanted.
* regressionwindow-wanted, testcase: regression from earlier versions is present, and good STR have been identified
I don't think we should change "testcase" to also mean "good STR have been identified". Again, we have a keyword for the latter: reproducible. (testcase implies reproducible of course) As I see it, a testcase is different from good STR. It's usually a small standalone file attached on the bug that reproduces the problem. So I think we already have keywords to represent the states you want. * regressionwindow-wanted: regression from earlier versions is present, but there are not clear, reproducible steps to follow (likely quite complex/intermittent) * regressionwindow-wanted, steps-wanted: regression from earlier versions is present, and some help crafting a testcase/clear steps to consistently reproduce is desired * regressionwindow-wanted, reproducible: regression from earlier versions is present, and good STR have been identified * regressionwindow-wanted, reproducible, testcase-wanted: as above, but we still want help producing a testcase from the STR * regressionwindow-wanted, testcase: regression from earlier versions is present, and we have a testcase that reproduces the problem where the last three are variations of the state you labeled as "good STR have been identified".
so I propose being more explicit about this.
Yes, I fully agree. And that goes for any bug, not just regressions. Cheers, Mats _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform