Where do you like bugs reported?
In my last post, a meta issue for modules: bug tracking, I had noticed a problem with the bug tracking link for a module and discussed that problem. In the comments, one person said he preferred rt.cpan.org. I began thinking about where to have bugs tracked for my modules. Since I have not published one yet, this is something I would like to know. I would like to know the good and bad and ugly of the various systems to make a more educated choice on issue tracking before my first release.
Are there specific issues with GitHub's, GitLab's, or other issue tracking systems making rt.cpan.org the more attractive choice?
On a side note, I prefer reporting issues on sites like GitHub and GitLab since my reply email is hidden and does not get spammed, or at least not yet. However, my cpan.org email address gets a lot of spam, so much spam I had to make a rule to send all email I receive through that address to junk mail. So, should I receive a reply to an issue I opened on rt.cpan, I may miss it since it ends up in my junk mail, which I do not check that often.
Where do you like bugs reported and why?
I'm an amateur HTML and Perl coder.
I read fantasy, watch science fiction, and listen to 80s music.
The reason I like rt.cpan.org is because it has incredibly flexible reporting on issues. Like I have my RT dashboard set to show six sections:
And for any of the sections, I can click on the heading to see the full list, beyond the top ten.
With Github issues, there's notifications, but once you've read the notification, it's marked as read and no longer so visible unless you go visit the issue tracker for that repository. And I have a lot of repositories. This can make it easy to lose track of issues.
RT gives me good ways of seeing an overview of issues.
You can get overviews of your issues across repositories on GitHub too. I’ve written about the ones I use. Of course queries at GitHub are relatively blunt as the available metadata isn’t nearly as rich as in RT. You can overcome that to some degree with elaborate label taxonomies, but you can only go so far that way. OTOH the orders of magnitude greater power available in RT comes at the cost of a certain… clunkiness. It would be nice to have an option somewhere in-between those extremes.
I prefer RT, because it's a permanent location.
My code hosting may move (e.g. bitbucket stopped supporting mercurial, so I had to move my code elsewhere). Moving issues between hosting providers is lossy and messy.
If my code repositories ever disappear, users still have a place to file bug reports so that future users or maintainers understand limitations of the package.