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

Re: The Serious severity

On Thu, May 02, 2002 at 03:10:34PM -0500, Manoj Srivastava wrote:
> 	Strawman.

?  I don't see how.

>  The rationale I presented argues for creating a severity to use for
>  violations of policy.  The point was to allow for violations of must
>  directives to be flagged as problems in themselves, potentially
>  release critical, thereby acknowledging the importance of Debian
>  policy for packages (IMHO policy is the major difference between the
>  solidity of a Debian machine vs other distributions).

The same function can be served by a tag, as the IRC discussion

>  Branden> Since you want to drag this out in the public forum of debian-policy,
>  Branden> I'll post some relevant hunks of IRC log.
> 	Talk about hypocrisy.  This is the same person who ranted
>  about lack of transparecy when he was not in the innner circle, but
>  has a problem with the DPL, RM, and a DPL candidate discussing a
>  corner stone of Debian like the BTS being dragged into the light of
>  day. 

Talk about leaping to conclusions.  I was simply waiting to come forward
with talk of a "consensus" between Anothony and myself until he had
signed off on it, so that I could be sure I wasn't misrepresenting his
position.  I am attaching the mail I sent him, so please feel free to go
ahead and accuse me of forging the Date: header before digitally signing

> 	Pardon me for trying to bring the people who do not IRC into
>  the conversation.

Pardon me for attempting to be courteous to Anthony by not speaking for

And, for the sake of openness and transparency:

03:38PM|* Manoj goes off to bait Overfiend on the policy list
03:39PM|<Manoj> serves him right for discussing serious severity minutes after I signed off to go to bed

I don't see how that attitude helps anyone.  Surely people do not need your
permission to discuss the utility of the "serious" severity?

G. Branden Robinson                |    If a man ate a pound of pasta and a
Debian GNU/Linux                   |    pound of antipasto, would they
branden@debian.org                 |    cancel out, leaving him still
http://people.debian.org/~branden/ |    hungry?              -- Scott Adams
From branden@debian.org Thu May  2 03:54:25 2002
Date: Thu, 2 May 2002 03:54:25 -0500
From: Branden Robinson <branden@debian.org>
To: ajt@debian.org
Subject: Anthony and Branden's consensus
Message-ID: <20020502085425.GQ23121@deadbeast.net>
Mime-Version: 1.0
Content-Type: multipart/signed; micalg=pgp-sha1;
	protocol="application/pgp-signature"; boundary="N7HXVILz59yg1nI8"
Content-Disposition: inline
User-Agent: Mutt/1.3.28i
Mail-Copies-To: nobody
X-No-CC: I subscribe to this list; do not CC me on replies.
Status: RO
Content-Length: 10227
Lines: 199

Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Any comments on this?  I'll put it into digestible form tomorrow, mail you
again to make sure I don't screw it up, and then pass along the resulting
document to the -ctte.

02:53AM|<Overfiend> I still do not see what harm would be done by shifting =
the first half of the "serious" severity definition to a tag
02:53AM|<Overfiend> certainly no automated tools would be adversely affecte=
d, aside from having to account for the change.
02:53AM|<bdale> Overfiend: so, propose doing so after woody releases
02:53AM|<Overfiend> bdale: where did I propose doing so before it releases?
02:53AM|<nwp_> IMHO the bugscan overrides are logically equivalent to such =
a tag, but less transparent.
02:54AM|* Shadur curses once more the S3 Savage series
02:54AM|<bdale> Overfiend: rephrase.  so, shut up about it until woody rele=
02:54AM|<Overfiend> nwp: ooh, transparency, once of my hobby-horses, as iwj=
 would put it
02:54AM|<aj> Overfiend: what harm would come of you calling yourself "Dubbl=
ebub" from now on? Not really any, it'd just be a nusiance while people got=
 used to the new description. That isn't really
             the question, the real question is what's the benefit.
02:54AM|<Overfiend> bdale: if people keep dialoguing with me about it, I'll=
 keep answering them.
02:55AM|<bdale> Overfiend: yes, I'm painfully aware of that.
02:55AM|<bdale> oh well, I still get more spam emails per day than OF email=
s, so it's not really a problem.  :-)
02:56AM|<aj> Overfiend: it would've been much better to have done the [IGNO=
RE] stuff from a month or two ago, it would probably have been good to have=
 [IGNORE] put in the BTS rather than in
             ~wakkerma/bugscan/comments on master, but those aren't going t=
o be fixed for woody (because no one but me is going to do anything about t=
hem; and i don't have the time to do them)
02:56AM|<Overfiend> aj: the benefit is that 1) package maintainers get to p=
reserve the pre-serious utility of their bug list as a triage tool; 2) the =
release manager gets to discern violation of
                    musts/requireds in policy 3) perhaps, the policy team h=
as an easier time of discerning compliance with certain policies, though th=
is could only really be realized with increased
                    adoption of the "Justification:" header
02:56AM|<aj> http://lists.debian.org/debian-devel-announce/2000/debian-deve=
02:57AM|<aj> is the list of bugs that were known about, considered "release=
-critical" and yet ignored anyway for potato
02:57AM|<jab> Why not just do what I did with FHS bug 143972 - mark it "nor=
mal" instead of "serious" ?
02:57AM|<Overfiend> jab: I tried that, aj and Manoj came down on me pretty =
02:57AM|<Overfiend> oh shit, now YOU'RE going to be in the doghouse, too :)
02:57AM|<jab> Overfiend: So what? You're the maintainer?
02:58AM|* Overfiend laughs at jab's na=EFvet=E9
02:58AM|* nwp_ gets the impression jab hasn't been following this one ;)
02:58AM|<aj> 143972 is a lot less blatantly wrong than the xutils bug
02:58AM|<Overfiend> jab: please read the bug logs of 97671
02:59AM|<Overfiend> aj: uh, the definition of "serious" says nothing about =
"blatantly wrong" versus regular "wrong", and I don't think the FHS makes s=
uch a distinction either
02:59AM|<Overfiend> I thought we wanted objectivity?!?
02:59AM|<bdale> aj: someday, you're going to have to let me in on the secre=
t of how you manage to so quickly and effectively find and emit url's for a=
ncient email messages of interest...
03:00AM|<jab> nwp_: Wow, some people really enjoy arguing. ;)
03:00AM|<aj> bdale: the release critical bug list is posted every few days =
to -devel-announce, in potato's time i was maintaining [IGNORE] info well, =
that message is at around the same time potato
             was released so should be the most canonical

03:30AM|<Overfiend> aj: okay, I gotta hit the sack soon.  Acknowledging tha=
t this is all post-woody stuff, what have we reached consensus on?  We shou=
ld mail the TC and let them know so they don't
                    fight the wrong battles for us.
03:32AM|<aj> Overfiend: we've reached a tentative consensus on declaring "s=
erious" nothing more or less than the RM's prerogative, that it might be wo=
rth making it be a tag...
03:32AM|<aj> hrm
03:32AM|<aj> surely the release manager tagging something "unreleasable" th=
en saying "oh, but it'll release anyway" would be even more annoying?
03:32AM|<Overfiend> Manoj will be unhappy, but he'll be especially unhappy =
if we leave no role for policy violations.
03:32AM|<Overfiend> or are you willing to deal with him on that front?
03:33AM|* nwp_ is disappointed to have missed the once-in-a-lifetime opport=
unity to see OF & aj reaching consensus while his network connection was do=
wn ;)
03:33AM|<Overfiend> there are a couple of ways we could go
03:33AM|<Overfiend> I would like to have serious-policy-violations represen=
ted by a tag, because of criterion 3)
03:33AM|<aj> policy violations can have a tag if he insists, but i've been =
planning on ripping the serious <-> must thing out for ages anyway. i have =
to repeat the "no, it's not like the RFC's"
             argument waaaaaay to often. you saw iwj come up with it again =
just recently, right?
03:34AM|<Overfiend> aj: yes
03:34AM|<Overfiend> 03:32AM|<aj> surely the release manager tagging somethi=
ng "unreleasable" then saying "oh, but it'll release anyway" would be even =
more annoying?
03:34AM|<Overfiend> Re: that, I don't think so
03:34AM|<aj> nwp_: it happens ocassionally. what'll be truly remarkable if =
we start out disagreeing and move to agreement *without* the horrific flame=
war in between
03:34AM|<Overfiend> if "unreleasable" is the RM's pissing ground, then peop=
le can be expected to guess what it means if the package releases anyway
03:34AM|<Overfiend> ideally we'd have a gizmo that auto-retagged them, but =
that's cosmetic
03:35AM|<Overfiend> and ideally it would hook into bugscan, etc.
03:35AM|*#* Zomb (blochedu@aixd1.rhrk.uni-kl.de) has joined channel #debian=
03:35AM|<aj> hrm
03:35AM|<nwp_> heh... well, good to see anyway. It's *so* fucking frustrati=
ng watching you guys violently agree with each other when you both want the=
 same thing really ;)
03:35AM|<Overfiend> in my conception, I as maintainer could tag a bug as un=
releasable if I felt it needed the RM's attention
03:36AM|*#* jannic (~jan@flaxian.hitnet.RWTH-Aachen.DE) has joined channel =
03:36AM|<Overfiend> i.e., "Gosh, this bug is pretty sucky, what do you thin=
03:36AM|*#* Signoff: ltd_ (Connection timed out)
03:36AM|<aj> that's the problem with tags, they'll probably tend to have to=
 be added after the fact, which is much harder to do than removing/downgrad=
ing after the fact
03:36AM|<Overfiend> but the RM's decision is final if he removed the tag, b=
arring further information
03:36AM|<aj> Overfiend: as a maintainer you get to declare your packages un=
releasable for whatever reason you choose
03:36AM|<Overfiend> oh, really?
03:36AM|<Overfiend> okay, that leaves a role for "serious", then
03:36AM|<aj> Overfiend: check the second half of the "serious" def'n
03:36AM|<Overfiend> as a severity
03:36AM|* ilm is packaging java2latex and wonders where to stick the .sty f=
03:36AM|<Overfiend> yes
03:36AM|<Overfiend> I wasn't sure you wanted to try eliminating serious or =
03:37AM|<ilm> would /usr/share/texmf/tex/latex/java2latex/ be a good place?=
 the package uses /usr/TeX/inputs/
03:37AM|<aj> (i made up "serious", all the stuff it covers is meant to be t=
hat way and works fairly well to a first approximation)
03:38AM|<Overfiend> aj: I'd like to avoid a retread of the current flamewar=
 by defining domains of authority.  the "serious" severity is for the maint=
ainer, the "unreleasable" tag would be for the
                    RM, and "serious-policy-violation" or whatever would be=
 for the Policy czars
03:39AM|<Overfiend> that way, whether the RM chooses to respect a given cri=
tical/grave/serious bug as truly RC would be up to him
03:39AM|*#* asuffield (aps100@ has joined channel #debian-dev=
03:39AM|<Overfiend> IOW I could say "oh my God, it would humiliate me if XF=
ree86 shipped this way, severity 12345 serious . tag 12345 unreleaseable", =
but the release manager can remove that tag and
                    say "tough, it's shipping"
03:40AM|<Overfiend> aj: is this consistent with what you understood our con=
sensus to be?
03:40AM|<Overfiend> and whatever the policy manual says is actually decoupl=
ed from BOTH of these concerns
03:42AM|<Overfiend> oh darn, did I lose him?
03:42AM|*#* Signoff: mhp (Connection timed out)
03:42AM|<tore> e
03:43AM|<tore> eh.
03:44AM|*W* aj is aj@azure.humbug.org.au (Anthony Towns)
03:44AM|*W* On channel #debian-devel
03:44AM|*W* On IRC via server irc.openprojects.net (http://www.openprojects=
03:44AM|*W* aj has been idle for 6:42; on since Wed Apr 24 00:43:12 2002.
03:44AM|<Overfiend> 03:44AM|*W* aj has been idle for 6:42; on since Wed Apr=
 24 00:43:12 2002.
03:44AM|<Overfiend> yup, I think I lost him :)
03:44AM|<Overfiend> oh well
03:44AM|* Overfiend archives this so he doesn't forget it
03:44AM|<bdale> email him a copy
03:45AM|<Overfiend> yeah
03:45AM|<Overfiend> bdale: see? you're so pessimistic
03:45AM|<bdale> Overfiend: no, I just think you could achieve the same resu=
lts more quickly and easily other ways.  [shrug]

G. Branden Robinson                |       The key to being a Southern
Debian GNU/Linux                   |       Baptist: It ain't a sin if you
branden@debian.org                 |       don't get caught.
http://people.debian.org/~branden/ |       -- Anthony Davidson

Content-Type: application/pgp-signature
Content-Disposition: inline

Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org



Attachment: pgpfDQLGRcMEg.pgp
Description: PGP signature

Reply to: