launchpad-dev team mailing list archive

[Benji York <benji.york@xxxxxxxxxxxxx>]

On Tue, Sep 20, 2011 at 9:22 AM, Stuart Bishop
<stuart.bishop@xxxxxxxxxxxxx> wrote:
> On Tue, Sep 20, 2011 at 6:52 PM, William Grant
> <william.grant@xxxxxxxxxxxxx> wrote:
>> On 20/09/11 21:49, Jeroen Vermeulen wrote:
>>> On 2011-09-20 18:04, Stuart Bishop wrote:
>>>
>>>> This ties into our 256+ critical bugs too. I'd really like to see our
>>>> high bugs downgraded to medium and our critical bugs downgraded to
>>>> high. This way we can use critical for the stuff that genuinely has to
>>>> be fixed right now possibly late at night and on weekends, unlike the
>>>> bulk of our existing critical bugs which we hope to deal with in the
>>>> next 6 months. 6 months away doesn't match most peoples definition of
>>>> critical...
>>>
>>> I agree.  We can't keep creating more Critical bugs than we can handle,
>>
>> Or perhaps we can't keep handling fewer Critical bugs than we are creating?
>
> There was a formula from an agile seminar I can never remember where
> you take the average time to fix bugs, rate of incoming bugs and end
> up with a timeframe. Any bugs hanging around longer than this
> timeframe are WONTFIX by definition, because the incoming rate of more
> important bugs multiplied by your velocity means you will never get
> around to it. So when a bug gets into that list it means one of a) Its
> been badly prioritized and you dropped it on the floor b) It will only
> be fixed by accident, such as becoming irrelevant or c) You're
> screwed.
>
> Or something like that. Anyone remember? I think my notes are buried
> in a box somewhere (Damn paper notes! How last millennium!)

It might be an application of Little's Law:  The average number of
customers in a queue is equal to the average rate they join the queue
multiplied by the average time each customer spends waiting in the queue
and being serviced (L = λW).

To apply it to bugs: if we have 10 new critical bugs a week on average
and each bug spends 3 weeks waiting and 1 week being fixed on average,
then we should expect to average 40 critical bugs open.  If you decide
that there is a deadline of 4 weeks to fix a critical bug, then if your
fix rate goes down or your incoming bug rate goes up, something will
have to give.

Sidebar: The law is more interesting when you only have L (average
number in the system) and λ (rate of additions to the queue) and want to
know W (how long from the time they enter the system to exiting the
system).  For example, if we know we average 10 new bugs a week and we
know that we average 250 open bugs, then we know that a bug will, on
average, be open for 25 weeks.  Of course the averages get less and less
useful the greater the variance of W.
-- 
Benji York