Intrepid Kernel Bug Stats

[Leann Ogasawara]

Hi All,

I thought it would be interesting to share some of the bug metrics I saw
for the Intrepid release and additionally how they compared to what we
saw for Hardy's release.

The following data points were taken for Hardy on April 24, 2008 and for
Intrepid they were taken on Oct 30, 2008.

Bug Status	# Intrepid Bugs 	# Hardy Bugs
==========	===============		============
New		950			97
Incomplete	1907			436
Invalid		746			153
Won't Fix	252			34
Confirmed	417			93
Triaged		518			295
In Progress	26			13
Fix Committed	49			13
Fix Released	1072			306
Totals		5937			1440

I'd like to point out some of the more interesting stats.  Generally
looking at the overall numbers, Intrepid saw a big increase in the
number of bugs in an open state (New, Incomplete, Confirmed, Triaged, In
Progress, Fix Committed) against the 'linux' kernel package.  I
attribute this inflation due to the kernel bug migration we did during
Intrepid.  Moving open kernel bugs reported against the older
linux-source-2.6.xx kernel packages to the current 'linux' kernel
package definitely boosted the number of bugs in an open state.

I'd also like to point out for some of the closed bug states (Invalid,
Won't Fix, and Fix Released) there is overlap with the Hardy numbers.
For example the 306 Fix Released bugs for Hardy are also included in the
1072 bugs Fix Released for Intrepid.  To get a better idea of the bugs
strictly marked Fix Released, Won't Fix and Invalid for Intrepid I did
the following:

Intrepid Fix Released = 1072-306 = 766 
Intrepid Won't Fix = 252-34 = 218
Intrepid Invalid = 746-153 = 593

It's interesting to see we marked 766 bugs Fix Released during Intrepid
vs only 306 for Hardy.  That's over twice as many bugs!  Additionally
218 Intrepid bugs were marked Won't Fix vs only 34 for Hardy.  And 593
bugs were marked Invalid vs 153 for Hardy.

The main factor that I think we can attribute the increase in the number
of bugs we closed has to do with the 2.6.27 call for testing we issued
during the Intrepid release cycle.  After having migrated the older
kernel bugs forward to the linux package coupled with the fact that
every linux bug was then issued the 2.6.27 call for testing, we received
a lot of feedback about bugs being resolved and that the reporter had
just plain forgot to close their report.

There is still a large volume of New and Incomplete bugs open that I'd
like to address.  To help explain the 950 new bugs, we didn't change the
state from New to Incomplete when we both migrated the older kernel bugs
as well as when we issued the 2.6.27 call for testing.  I believe a
majority of them have not been tested against Intrepid and should be set
to Incomplete.  Also, many of the 1907 Incomplete bugs should likely be
closed because we haven't received feedback about testing with Intrepid.
However, we've left the closing of the Incomplete bugs to be handled by
the Launchpad janitor whenever it eventually gets turned on.  

I made some pie charts to help visualize the data:

http://people.ubuntu.com/~ogasawara/bugdata/intrepid-linux-total.svg
http://people.ubuntu.com/~ogasawara/bugdata/hardy-linux-total.svg

I also made some charts strictly for the bugs that made the weekly
kernel bug lists I send out.  I feel these weekly bug lists are still
proving to be helpful as it looks like over half of the bugs I give you
get closed.

http://people.ubuntu.com/~ogasawara/bugdata/intrepid-ogasawara-linux-buglist.svg
http://people.ubuntu.com/~ogasawara/bugdata/hardy-linux-ogasawara-buglist.svg

Anyways, just thought I'd share the numbers and how I interpreted them.

Thanks,
Leann