Chris Lawrence wrote: > My gut feeling is that we let the users do the first level of bug > triage, and then developers can agonize about reclassifying the < 5% > they get wrong. I'd rather see 100x too many grave reports than let > one slip through at "normal" because we put some artificial barrier in > the way of bug reporting.
That 5% is an interesting number. I think it's close to right, when you look at all the bugs; most users get wishlist and normal severities right, and I rarely get an incorrect important severity bug anymore. However, when you look at the RC bugs, I've been keeping track for 2 or 3 months now of how many RC bugs I've gotten where it was not inappropriate to immediatly issue a 'bts severity xxx normal' (and often 'bts tag xxx unreproducible'). Of quite a few RC bugs in that time frame -- probably 20 or so -- there were just two where it was not appropriate to do that, and both were filed by debian developers. Maybe that 5% is coming entirely from the RC bug severities.. > (I also don't know how much Branden's Justification patch will address > your concerns; it's only been in testing for a week or so... but in > theory it should cut down on this problem.) I hope. -- see shy jo, who has considered a procmail recipe
