On Tue, Oct 07, 2008 at 08:11:51AM -0400, Ricardo SIGNES wrote: > > Yesterday I released my "make dist" replacement, Dist::Zilla. Later, > I went to http://www.cpantesters.org/show/Dist-Zilla.html and saw the > test reports for it. I was pretty surprised to see results for > pre-1.00 dists. ..... > > So, a report sent in several months before the dist was released was > indexed and stored. This seems potentially slightly troublesome, as > people may do the same thing I did, but with code they don't mean to > divulge the existence of, or may otherwise screw up reports just > enough to be useless.
Don't do that then ;^) > Anyway, I'm mostly reporting this because it's weird and maybe > unintended. I could also believe if they're accepted on purpose to > avoid races -- although four months is quite a race. No it's always been that way. Accept whatever comes in and parse it. In the short term I'm not going to change it, as it's partly down to the author to do not that, and the fact there are higher priority fixes on my TODO list. In the longer term I can perhaps add a check to see whether the distribution has ever been submitted to CPAN. The code to determine that is in the Reports mechanism now, so it won't be too traumatic to port it over to the data generation code and automatically ignore it. To coincide with this I'm looking at a mechanism to mark reports for deletion. Or rather set an ignore flag, for those reports that are deemed bogus by both the author and the tester. In this case they would both be you, so you should easily set the ignore flag. The report pages for that author/distribution can then be rebuilt to reflect the change. Where a tester is unresponsive, an admin (probably xdg, binGOs and myself) should also be able to set the ignore flag. Cheers, Barbie. -- Birmingham Perl Mongers <http://birmingham.pm.org> Memoirs Of A Roadie <http://barbie.missbarbell.co.uk>
