2007's GSoC ideas for Gentoo had this:
"Improve upon Gentoo's Bugzilla (either within bugzilla or externally) to get better metrics per CP(V). Have some sort of AJAX lookup for real CPV's. Store the CPV per bug (how to do bugs w/multiple CPVs?). This lets us harvest bug information better because we can bind CPV's to bugs." (from
the userrel/soc page)
As an AMD64 AT, I know that quite often things on
Gentoo's Bugzilla don't run as smoothly as they should. However, the above proposal seemed difficult to integrate with the existing Bugzilla. My proposal is to write something new from scratch that merges the functionality of
packages.gentoo.org and
bugs.gentoo.org into a state-of-the-art web application.
Pros
- Reduced bugspam, traffic and server load
- Significantly reduced development overhead
Cons
- Lots of work
- Difficult transition period
Whenever I had nothing else to do in the last few months, I tought myself
django and worked on my latest pet app:
coisy.
And here's the good news: coisy already does most of what packages.g.o does and has the foundation for the integrated bugtracker. Here are some screenshots of my current progress (note that the layout and everything is nowhere near final, this is just supposed to communicate that something is actually working here):
So here's a new list, with respect to coisy:
Pros
- Reduced bugspam, traffic and server load
- Significantly reduced development overhead
- django prevents the vast majority of the SQL injection or XSS stuff you'd have to worry about
- coisy can offer detailed statistics to evaluate the state of development
- coisy scales nicely
- coisy allows developers to stay up-to-date using RSS feeds, resulting in a less cluttered inbox (email notifications will still be available, of course)
- coisy can track any number of trees (read: overlays) and manage them independently (e.g. a bugtracker for sunrise or whatever)
Cons
Lots of work half of it is done, I'll do the rest
- Difficult transition period
Transition from Bugzilla to coisy
First of all, I don't see this happening within the next 12 months at all (probably longer). I am aware of that most developers will be
very reluctant about this change, so I came up with some ideas to handle the transition. Gentoo's Bugzilla contains more than 210,000 bugs, containing a wealth of important information. Therefore, it should remain readable as long as possible. When deploying coisy, I suggest leaving bugs.gentoo.org in readonly-mode. Coisy will have the possibility of tying a coisy ticket to a Bugzilla bug and to automatically import comments and attachments from a given old bug.
My goals for this Summer of Code
At the end of SoC, coisy won't be ready for production. My goal is to complete packages.g.o functionality and get some rudimentary bugtracking going.
Beyond Summer of Code
For the time when SoC is over, I still have many ideas for coisy. How about just replying to bugmail in order to post a comment?
Why I want to do this as a SoC project
So far, I've been hacking on coisy whenever I had nothing else to do, with sometimes several weeks without a commit. And I will continue to do so even if my proposal gets rejected. However, as a SoC project, coisy would go on top of my TODO list and I'd have much more motivation. Also, I expect coisy to gain some exposure through SoC, which hopefully means some community and developer input for me.
Why I think Gentoo should accept my proposal
Simple: There's no risk for Gentoo. Even if coisy isn't transformed into a GLEP and never goes live, Gentoo hasn't really invested any resources into its development. However, if it goes well, Gentoo would have the funkiest development platform on the block... or whatever.