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

"Release-critical" bugs (was: Re: Processed: yawn)

On Sat, Apr 20, 2002 at 09:57:12PM -0500, Manoj Srivastava wrote:
>         It is my considered optinion that this is a serious
>  bug. Whether or not the bug is release critical depends purely on the
>  release manager. 

It seems some clarification about this might be helpful.

For those tuning in late, the context is in bug #97671. It's not
particularly needed to grok the rest of this mail, though.

Critical, grave and serious bugs are intended to be tied to particular
types of errors and to be relatively objective.

Further, there's intended to be a one-one correspondence between issues
matching the criteria for critical, grave, and serious and issues that
make packages unsuitable for release with Debian.

Most of the time how this is dealt with is relatively straightforward:
effort and attention is focussed on fixing it, or, if the bug is too
hard and the package doesn't matter much, it's dropped.

When the release gets down to a couple of weeks away, though, this
attitude changes, and some classes of bugs that are considered critical,
grave, or serious get set aside rather than dealt with. As woody's release
approaches, eg, the following bugs are being ignored in this manner:

     * #144330: gcl: FTBFS: tk-dev does not exist (hppa/unstable)   
     * #144380: gnotepad+ has typo in Conflicts: line
     * #141922: ldirectord: lvs is renamed to ipvsadm 
     * #144299: libwmf0 and libwmf-dev have file overlap
     * #139689: mesag-dev: version install fails (conflicts
       with mesag-widgets-dev) on ppc
     * #144254: setserial: maintainer scripts modify conffile
     * #144255: squid: maintainer scripts modify conffile /etc/squid.conf
     * #129819: syslog-ng: klogd stops working after syslog-ng is
       reloaded or restarted
     * #137752: xfree86v3 fails to build from source on powerpc
     * #139523: building on arm (and probably others) fails
     * #143825: xutils: why is rstart.real a conffile?

The argument for this way of doing things is pretty straightforward,
and you may've seen it before. A set of severities is need to precisely
identify release-critical issues so that they can be appropriately
dealt with -- nothing else is feasible. These severities need to be as
objectively determined as possible, as we have far too many bugs to run
them all past an oracle to decide what severity they should be. We need to
fix or remove packages with these problems so that, generally speaking,
our releases meet our standards. And, finally, we need to have special
exemptions because, quite frankly, our standards are marginally higher
than our abilities.

For comparison with how this was working with potato, you might like to
look at:


Unfortunately there's no similarly annotated list of the RC bugs
affecting woody.


Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.

     ``BAM! Science triumphs again!'' 
                    -- http://www.angryflower.com/vegeta.gif

Attachment: pgplifX05JGAs.pgp
Description: PGP signature

Reply to: