>>>>> On Wed, 25 Nov 2009 12:10:55 -0500, David Golden <xda...@gmail.com> said:
> I was thinking about Tim's email more and want to add that I think it > is worth it to have more testers submitting reports even for common > platforms. We're nowhere near having statistically valid samples on > most distributions on common platforms. If we get to the point where > we have 10 or 15 reports from platform X with identical config > settings, then I think we can start encouraging those testers to > diversify more. I've also come to Tim's conclusion that we already now need more diversification than we have. But it's less a matter of encouragement but a question how to support it. One problem I see at the moment is the format of the reports-sent.db file: test FAIL Data-FormValidator-4.63 (perl-5.11.2) x86_64-linux 2.6.30-1-amd64 Right now a reporter will send a maximum of one pass and one fail for this distro/perl combination. Imagine a bughunter is trying out variations in the C compiler or in the toolchain or in installed versions of modules required or is bisecting perl. Then it would be wonderful to see the resulting variations in cpantesters. Currently the bughunter has to wipe out the one line in the reports-sent database to deliver more than one PASS or FAIL. What would you think about a report_again command? -- andreas
