[dtp-dev] Countdown to DTP 1.6M6 and bug density

[jograham@xxxxxxxxxx]

DTP 1.6M6 is only 14 working days away, and there are far too many bugs
assigned to the "1.6M6" target milestone. For example, yesterday we had 52
and today 53, which translates into a bug density of 3.47 and 3.79.
Essentially this means that we'll have to resolve 4 bugs per day from now
until 3/28 to clear the M6 target milestone list.

We are very, very unlikely to accomplish this because:

* The counts are only for current bugs, and do not include incoming.
Typically in each milestone there are a number of incoming bugs that must
be addressed for that milestone, and which take a substantial amount of
time to resolve, test. and verify.

* We have never been able to sustain a closure rate of more than 2 bugs per
day. Sure, there have been days. or even a couple of days in a row, when
more than 2 bugs per day were closed, but those periods are rare and there
is no reason to believe that such sprints can be sustained over three
weeks.

* EclipseCon is next week, and thus a number of DTP committers will not be
available to resolve bugs during the conference

We must be very careful to avoid two situations that can arise when the bug
density gets to an impossible level:

* We must not resolve a large number of bugs quickly right at the end of
the milestone cycle, since doing so greatly increases the risk of
additional defect injection. Simply put, rush work at the end of an
iteration is rarely done well, while often at the expense of breaking
builds and creating more bugs in the process.

* We should not wait until the last moment to postpone (reassign) a set of
bugs, since adopters depending on DTP might be impacted by such delays. If
we instead start reassigning bugs out of M6 now, then adopters will have a
reasonable period of time to react and perhaps petition us for
reconsideration if a particular bug really is critical.

So, I'd like to ask every DTP committer to look at their 1.6M6 target
milestone list as soon as possible and make the necessary adjustments. From
my perspective a bug density of around 1.5 bugs per day is reasonable.

Regards,
John Graham
Eclipse Data Tools Platform PMC Chair
Staff Software Engineer, Sybase, Inc.
http://osmusings.blogspot.com/
http://dataplat.blogspot.com/