[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Debian 2.2 Release.



Tomasz Wegrzanowski wrote:
> If upstream says true that app is stable, 99% bugs can be detected
> by lintian, apt-get install newapp, try it once, apt-get remove newapp.

Let's look at all open release critical bugs, shall we? I've categorized
them as detectable by your simple testing methods above, or not. If I
had any uncertianty, I assumed your test would get them. I even assumed
your test really means install it and test it on all architectures
debian supports.

Bug number	Simple test?	Description
#64371		no		Hardware-specific problem.
#64500		yes		Install problem on ppc
#65515		yes		Problem passibg parameters to modules.
#65988		yes		Symlink problem breaks install.
#66284		yes		PCMCIA support loaded twice.
#65800		yes		Architecture-dependant coding error.
#65644		yes		Library incompatability.
#65052		no		Copyright problem.
#65051		no		Lacks libc6 dependancy.
#65454		no		Doesn't work with emacs20 version mismatch
#65208		no		Incompatability with new emacs
#66267		yes		Compile problem on ppc
#66172		no		Causes emacs to not load some modules
#66215		yes		Doesn't start window manager
#65962		no		Locks some hardware
#65967		no		Incompatability with X on some systems
#62173		no		Copyright problem.
#66114		no		Security hole.
#65857		no		Session management problem.
#58367		no		Locks box sometimes
#54271		maybe		Dependancy problem.
#62363		no		One driver has a problem under heavy load.
#66222		no		Doesn't stop properly on reboot.
#64414		yes		Library incompatability.
#66290		yes		Not executable.
#66142		no		Breaks minicom, efax.
#64762		maybe		Dependancy problem.
#65961		no		License problems.
#66086		yes		Doesn't work with gnuchess.
#66216		yes		Doesn't work on alpha.
#66137		no		Fails to install in some circumstances.
#65834		no		Hardware-specific issue.
#66225		yes		Doesn't work after some upgrades.
#51747		no		Doesn't handle expired passwords.
#64424		no		Wrong path for xauth on ppc.

So your simplistic tests would catch, optimistically, 13 RC bugs, or 37%
(not the "99% bugs" you asserted with no evidence), while 20 RC bugs, or
57%, would not be caught.

Do you really feel like halfing the number of release critical bugs we
fix in the released version of Debian by crippling our testing process
to what you've proposed?

-- 
see shy jo, wondering why he's wasting his time on this



Reply to: