Ease of bug reporting == caring for the users?
Being on the #perlde, the German IRC channel I got into an argument about the difficulty in submitting bug reports to Padre. One of the comment went as far as saying: if he doesn't care about other people he should not publish his software.
Every module on CPAN automatically gets an RT Queue. You cannot opt-out of it. You will get it if you want it and if you don't. In a way this is a good thing as it means no matter how unresponsive the author is the users can always send their bug-reports to some place.
Then I heard that there is an author, at least one, who responds to every bug report opened on RT with a harsh messages that tells the user she should send the bug reports directly to him.
Then there is Padre which is using its separate Trac system for bug reports and feature requests. Not only that, but it also requires the reporter to register first and due to heavy spamming the reporter first needs to talk to the Padre developers on IRC. These extra steps are clearly reducing the number of bug reports and they might even frustrate some of the potential reporters. After all they just want to report something and get on with their life.
I can symphatize with that a lot. It happened to me many times that I fund a bug in a project and just wanted to report it without further involvement. To me that was enough contribution to the give open source project.
Unfortunately that can easily create lots of duplicate bug reports and lots of partially written reports and even many false reports. People can easily report problems that are due to some other issue, not necessarily a bug in the code.
There is a trade-off between ease of reporting bugs and the quality of bug reports. I am quite sure if we made it easier to open a bug report we would get a lot more of them. We see that people sometimes come to the Padre IRC channel and ask about a behavior they see. In many cases those turn out to be errors outside of Padre and can be fixed in a short discussion. If that was reported as a bug it would have frustrated the developers, created unnecessary work for them and probably the most important would have not solved the problem.
opt-out from RT
So how did Padre opt-out from having an RT Queue? It is simple. We bribed Jesse.
Once we found that it does not work, we learned about the option in META.yml
Setting
resources: bugtracker: http://padre.perlide.org/trac/
will make search.cpan.org display a link to our bug tracker instead of the default RT Queue. The RT queue is still there and is still active but most people will find our own bug tracking system and use that. So far I think we had one bug report there. We copied it to our trac system and mentioned it on th RT report. No big deal. You don't even need to have an account for this on RT.
Would it be good if every CPAN module had its own bug tracking system? Probably not.
So why do I think it is OK for Padre to have one?
Padre is probably one of the largest projects on CPAN though I am not really sure how that can be measured. It is certainly the buggiest. It has 479 open tickets.
Most other large Perl projects are not distributed via CPAN and thus this question was never asked.
Most modules that are on CPAN have a lot less open tickets. Is that because they are smaller in scope? Better tested? More mature? Unused? Maybe it is because it is harder to open a bug on RT for people who don't yet know how to do it?
BTW as I can see [dist://Dancer] also uses and external bug tracking system and it has 87 open issues.