minutes of ESC call ...

[Michael Meeks]

Hi Cor,

On Sun, 2012-06-24 at 23:23 +0200, Cor Nouws wrote:
> Michael Meeks wrote (21-06-12 17:55)
> > * 3.5 bugs tagged with 'regression'
> > 	+ 179(+12) bugs open of 631(+17) total
> > [...]
> 
> Just taking a quick look at issues from the query below, I would suggest 
>   to skip mentioning numbers. There is pollution with old issues,
> regressions in new functions, minor-if at all issues ..

	Sure - and there is double counting of some issues - as I point out in
the numbers - which gives a misleading spike recently.

	Furthermore, an increase in the apparent number of regressions may be
related to an increase in QA activity categorising and triaging bugs :-)
and thus totally un-related to any underlying improvement / degradation
in real quality.

> The query is very useful to track and look at the reports. But the 
> number gives a wrong idea of quality. It only shows there's a lot QA 
> /triage work pending. And that was not really news ;-)
> (Might well be that the same applies for other numbers..)

	True; OTOH, this is not a news headline, it is a motivational set of
numbers that are designed to try to encourage people to shrink the
numbers.

	Already (IMHO) they have helped to improve our component categorisation
(the writer guys got sick of being blamed for all common / core code
bugs), and (I hope) also encouraged people to work on getting the
numbers down. The hope is that by double counting crashers eg. we
encourage people to focus first on those.

	So - the numbers are by no means perfect, but hopefully they serve a
useful purpose and show at least something of our QA activity, and
developer / bug-fixing work.

	The overall graphs I produce also show bugs fixed, where you can see
(from the growing separation of the 'open' and 'closed' trends) that we
are fixing more most annoying & regression bugs over time, and
particularly Most Annoying Bugs :-)

	At least, that's my thinking :-) do you have some better insights ?

	All the best,

		Michael.

-- 
michael.meeks at suse.com  <><, Pseudo Engineer, itinerant idiot