Hi David, On 27/10/2007, David Golden <[EMAIL PROTECTED]> wrote: > On 10/27/07, Sagar R. Shah <[EMAIL PROTECTED]> wrote: > > I guess there's still no way for a module author to reject a test > > report as not being a module bug - i.e. the FAIL adds to the fails > > count for this distribution forever? > > No -- but please think of reports just as data. It's not Kwalitee or > anything like that. I had failing reports on CPAN::Reporter for ages > from one particular tester -- we finally figured out their > Module::Build install was missing a file for some reason. It's just > data from people testing in the wild -- perfection is not a realistic > goal. > > In this case -- I realize it's worse because the testing tool itself > caused problems and I apologize for that. It's why I rushed out a fix > at 6:50 in the morning, mere hours after you sent your email about the > fail reports you got. > > If you're particularly concerned about bogus fail reports, just bump > your version number and re-release and the smokers should crank out a > fresh bunch of pass reports in short order.
I think you've taken my comment a little more seriously than intended. (the eternal danger of email as a medium ;-) ) I do indeed think of the reports just as data and am aware there are many conditions that can create dud failures. At the same time I know that there also exist a great many uneducated Perl developers out there who still think of it as a scripting language and might misinterpret the counts. My comment was really an off the cuff remark/query rather than an attempt to imply there was something deeply wrong with the system or a complaint against you/CPAN::Reporter. Now I think about it for more than 10seconds the right thing of course is to educate those people rather than adding unnecessary functionality to the CPAN tests system. The speed at which you got the fix out really is appreciated. In another non-Perl community the fix might have taken months irrespective of the effort required ;-) Best Regards Sagar
