Various problems seen in RC1

[Robert Watson]

On Sat, 26 Sep 2009, Aristedes Maniatis wrote:

> I've written before about this, so I will not go into detail. And I notice 
> that Mark was funded by the Foundation to replace the bug tracker with 
> something else. Was there an outcome from that work? My volunteer to help 
> was not replied to, so I assume that others have the matter completely under 
> control, but it would be nice to see what direction this is heading in. Most 
> other open source projects I am associated with have much more sophisticated 
> bug trackers which tie in tightly to the development and release processes, 
> but for the 8 release cycle, the best we have is Robert's TODO wiki page 
> which is usually at least a week out of date. It is interesting to note that 
> on that page there are only a tiny number of references to PRs.

Unfortunately, I fell behind due to running the developer summit and 
EuroBSDCon, but really that's a symptom of an underlying problem: I was (and 
am) manually maintaining the wiki list based on re@ e-mail correspondence.

That is fundamentally the wrong approach--we should be using the bug-tracking 
system to manage pending requests and known issues.  In particular, I'd like 
each merge request and in-progress issue to be captured by a bug entry, and 
referenced during commits and merges.  This would allow 99% of the information 
on the wiki page to be mechanically generated, and avoid the "missed stuff" 
problem.  We'd also have to get better at saying "it's a real bug but we can't 
fix it for this release" explicitly, of course.

There's not an opportunity to fix that for 8.0, but my recommendation has been 
that we at least use gnats, if not some more capable, issue-tracking system to 
handle pending changes and merge requests, with approvals to commit to 
branches linked back to the request so re@ can track what's going on.

Robert N M Watson
Computer Laboratory
University of Cambridge