On Friday 05 September 2008 21.08.10 David Golden wrote: > For anyone still following the other threads, I plan to make changes > to Test::Reporter that will stop authors from being copied on reports.
Make the least useful behavior the default is, well ... !0#?, up to you but is a bit of non sense to me. I understand why you want to do that and I'd be happy to see those that do not want to receive mail notification stop getting the mails. The system works today. Authors not happy with the system should log to your centralized system (which I dislike, read below) and say they don't want mails not the other way around. The smoke test results are a good reminder, to new developers and not so new, that what they do is not just used by them. If they dislike getting results they can say "no thanks" as any of us but eliminating the mails, by default, eliminates one possibility to educate new module authors. The all-on or all-off solution is not good. Neither is the centralized solution. Having a central web page, database, ... is also going to be more work for you and it will need to be maintained (thank you for all the work you are already doing btw). Why not let authors decide which modules they want to get smoked? Why not let authors decide which platform they want their modules smoked on? It seems to me that the smoking parameters should be distributed to the modules and have a fine granularity. This information is part of the module meta data. Smoking _is_ important and has been an important factor in the success of Perl. Pleasing users is good but that shouldn't be done at the cost of loosing an important quality factor. These are the years of "entreprise Perl" and smoke tests _and_ the distribution of the information about smoking are important. If you still opt for the centralized control, will it be possible to install a private smoke environment and a private centralized control? Cheers, Nadim.
