From tim@x22  Fri Jul  4 04:37:04 1997
Received: from x22 (ppp1669.on.sympatico.ca [206.172.249.133])
          by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id EAA21942
          for <FreeBSD-gnats-submit@freebsd.org>; Fri, 4 Jul 1997 04:37:03 -0700 (PDT)
Received: (from tim@localhost) by x22 (8.8.5/8.7.3) id HAA00337; Fri, 4 Jul 1997 07:36:39 -0400 (EDT)
Message-Id: <199707041136.HAA00337@x22>
Date: Fri, 4 Jul 1997 07:36:39 -0400 (EDT)
From: Tim <hoek@hwcn.org>
Reply-To: ac199@hwcn.org
To: FreeBSD-gnats-submit@freebsd.org
Subject: gnats re-opening prs
X-Send-Pr-Version: 3.2

>Number:         4028
>Category:       misc
>Synopsis:       GNATS auto-magically re-opened 14 prs
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    steve
>State:          closed
>Quarter:
>Keywords:
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Fri Jul  4 04:40:00 PDT 1997
>Closed-Date:    Sat Aug 23 09:41:12 PDT 1997
>Last-Modified:  Sat Aug 23 16:00:01 PDT 1997
>Originator:     Tim Vanderhoek
>Release:        FreeBSD 2.2.2-RELEASE i386
>Organization:
Lost
>Environment:

FreeBSD.ORG

>Description:

There are 14 problem-reports which were closed at some point in the
past but, at some point, I believe within the last month, they lost
all of their audit-trail and their state switched to never.

I have a feeling that all 14 of the prs which re-opened were the ones
that were sent when hub was re-assigned, gnats backed-up, and when it
began working again, all prs were duplicated.


>How-To-Repeat:

I don't know.  But the fact that 14 prs suddenly lost all of their
audit-trail disturbs me somewhat, and I sure would like to know.


>Fix:

Well, it's probably easiest to just reclose the 14 prs.  I made this
list by snarfing a "Current Problem Reports" list that was sent to
freebsd-bugs and then comparing it to the pr-list that query-pr returns
now.

Here's a link to the newest list of prs that was mailed to freebsd-bugs
that I could find.

http://www.freebsd.org/cgi/search.cgi?words=current+and+problem+and+reports&max=250&source=freebsd-bugs&docnum=197

3421
3422
3423
3424
3425
3426
3427
3428
3430
3431
3432
3433
3434
3435
>Release-Note:
>Audit-Trail:

From: Bill Fenner <fenner@parc.xerox.com>
To: freebsd-gnats-submit@freebsd.org
Cc:  Subject: Re: misc/4028
Date: Sat, 5 Jul 1997 19:57:18 PDT

 In ignorance of this PR, I attempted to clean up the GNATS database.  I
 used what I thought was the traditional rule to close later duplicates
 and leave the earlier one open.  If the PR's listed in this PR were all
 really closed, then I closed the "wrong" PR's.
 
 A list of PR's that I closed follow.  If someone else wants to
 clean this list up in one fell swoop I'd be happy to undo the
 damage that I did.
 
 Closed:			Duplicate for:
 bin/3437		bin/3424 (which I later closed in response to
 				  http://www.freebsd.org/cgi/search.cgi?words=su+acting+strange&max=250&source=freebsd-bugs&docnum=1,
 				  which Tim Vanderhoek pointed out to me).
 gnu/3433
 kern/3423		kern/3427 (which I also closed)
 kern/3427		kern/3446
 kern/3447		kern/3428
 bin/3425		bin/3422 (which I also closed)
 bin/3422		bin/3451
 

From: hoek@hwcn.org
To: freebsd-gnats-submit@freebsd.org, fenner@parc.xerox.com
Cc:  Subject: Re: misc/4028 : GNATS auto-magically re-opened14prs
Date: Mon, 14 Jul 1997 17:45:21 -0400 (EDT)

 >
 >There are 14 problem-reports which were closed at some point in the
 >past but, at some point, I believe within the last month, they lost
 >all of their audit-trail and their state switched to never.
 
 After examining the freebsd-bugs archive, I can say a couple more things.
 
 1) The bracket during which they re-opened can be definitively placed
 somewhere between June 2 and June 30, inclusive. 
 
 2) I had suggested Joerg Wunsch that the audit-trail could be (relatively)
 easily reconstructed from the freebsd-bugs mailing archive.  That was
 a lie.  There does not appear to be any mention of the listed prs in the
 archive (except for Bill Fenner closing some and for their mention in this
 pr).  ie. The original pr and subsequent open-close message was either not
 posted to freebsd-bugs (for whatever reason) or they were posted, but not
 archived (for some reason).
 
 (I will not that I could not search the whole freebsd-bugs archive, since
 that thing is very difficult to ftp (it's not even gzipped!).  From byte
 ~29900000 to 30000000 have not been checked by me.  Checking this is left
 as an exercise to the reader with easier access to the mail archives
 (actually this whole thing should've been left as such...)).
 
 From 7:40 to 18:06 on April 30, no messages are archived in freebsd-bugs.
 Perhaps the prs were closed quickly and perhaps the freebsd-bugs archive
 was unwritable for a short period (this _was_ during the hub-freefall
 transition) and that is perhaps why there isn't even an open-closed
 state-change message for them.
 

From: Bill Fenner <fenner@parc.xerox.com>
To: hoek@hwcn.org
Cc: freebsd-gnats-submit@freebsd.org
Subject: Re: misc/4028 : GNATS auto-magically re-opened14prs 
Date: Tue, 15 Jul 1997 07:49:11 PDT

 If the original PR's were listed as confidential, nothing related to
 them would have been sent to the freebsd-bugs list.  If they were
 later (silently) changed to be non-confidential, as is the tradition,
 they would suddenly show up as you saw.
 
 Are there bug summaries that have them, then don't have them, then
 have them again, or just don't have them then do have them?
 
   Bill

From: Tim Vanderhoek <hoek@hwcn.org>
To: Bill Fenner <fenner@parc.xerox.com>
Cc: freebsd-gnats-submit@freebsd.org
Subject: Re: misc/4028 : GNATS auto-magically re-opened14prs 
Date: Tue, 15 Jul 1997 12:07:07 -0400 (EDT)

 On Tue, 15 Jul 1997, Bill Fenner wrote:
 
 > If the original PR's were listed as confidential, nothing related to
 > them would have been sent to the freebsd-bugs list.  If they were
 > later (silently) changed to be non-confidential, as is the tradition,
 > they would suddenly show up as you saw.
 
 I considered this, but decided it was impossible since I know for
 certain (ie. [100 + 10]% certain) that at least some of the
 affected were never at any point in time marked confidential. 
 
 (Unless they were silently marked confidential later.  Very very
 unlikely).
 
 Besides.  Even if they got changed from confidential to
 non-confidential, that still doesn't explain them re-opening
 (even if it explains why they can't be found on the bugs
 archive). 
 
 I don't know how gnats stores bugreports, but if it stores all
 changes to a pr as deltas to the original, and the deltas somehow
 got destroyed or disattached (from the original pr, leaving them
 floating around?), that would explain them re-opening. 
 
 
 --
 Outnumbered?  Maybe.  Outspoken?  Never!
 tIM...HOEk
 
 

From: Tim Vanderhoek <hoek@hwcn.org>
To: Bill Fenner <fenner@parc.xerox.com>
Cc: hoek@hwcn.org, freebsd-gnats-submit@freebsd.org
Subject: Re: misc/4028 : GNATS auto-magically re-opened14prs 
Date: Tue, 15 Jul 1997 16:40:08 -0400 (EDT)

 On Tue, 15 Jul 1997, Bill Fenner wrote:
 
 > Tim Vanderhoek <hoek@hwcn.org> wrote:
 > >Specifically, bugs which were closed became silently open.
 > 
 > Are you sure they were closed?  Or did they just not appear in the
 > bug summary because they were marked as confidential?
 
 Ok, if they were marked confidential, then this is one
 of the things that would have had to have happened.
 
 1) Bug report gets sent using send-pr.  Confidential is set to
 "no".
 
 2) Bug report arrives and gets stuck in queue that resulted from
 hub-freefall split.
 
 3) gnats handles pr.  Two things happen.
 	a) pr gets duplicated
 	b) Confidential on one of the duplicates gets set to
 "yes" and stays "no" on the other duplicate.
 
 I'm using gnu/3433 as an example here.  For some of the others,
 gnats would have had to change Confidential to "yes" on both
 duplicates.  Still others may have slight variations, but this is
 what would have _had_ to have happened.
 
 Further, Mike Pritchard (gnats-meister at the time) was aware
 that pr's got duplicated.  I would imagine he was looking for
 related possible problems and would have seen such a strange
 anamoly. 
 
 I'm not sure if this is more or less plausible than suggesting
 that gnats arbitrarily removed their audit-trail.  Third
 suggestions are welcome, but ultimately, I'm not sure how much it
 matters how it happened.  My original hope had been that they
 could be (fairly) easily pieced back together again, but without
 some other suggestion, I'm not sure that's possible, meaning it
 may be easier just to let the things sit and wait to be looked at
 again individually.
 
 
 > >The list of prs affected I constructed by comparing a known good
 > >pr-list posted to freebsd-bugs with the current list on
 > >www.freebsd.org.
 > 
 > Was there a previous pr-list with the bugs open?
 
 No.  There is about a 5 day gap during which they had to be
 closed in order not to get listed on the pr-list posted May 5
 (unless the arrival-date shown on the prs is way off).
 
 
 --
 Outnumbered?  Maybe.  Outspoken?  Never!
 tIM...HOEk
 
State-Changed-From-To: open->closed 
State-Changed-By: steve 
State-Changed-When: Sat Aug 23 09:41:12 PDT 1997 
State-Changed-Why:  
All affected PRs have been re-closed. 

From: Tim Vanderhoek <hoek@hwcn.org>
To: Steve Price <steve@freebsd.org>
Cc: steve@hub.freebsd.org, freebsd-bugs@hub.freebsd.org,
        freebsd-gnats-submit@freebsd.org
Subject: Re: misc/4028
Date: Sat, 23 Aug 1997 14:33:29 -0400 (EDT)

 On Sat, 23 Aug 1997, Steve Price wrote:
 
 > Synopsis: GNATS auto-magically re-opened 14 prs
 > 
 > State-Changed-From-To: open->closed
 > State-Changed-By: steve
 > State-Changed-When: Sat Aug 23 09:41:12 PDT 1997
 > State-Changed-Why: 
 > All affected PRs have been re-closed.
 
 Nope, you got one wrong.
 
 Either kern/3428 or kern/3447 should be _reopened_.
 
 Bill Fenner accidentally closed kern/3447 since it was a dup of
 kern/3428.  You (Steve Price) just closed kern/3428 because it
 was magically re-opened.  At no time was the actual bugreport
 addressed.
 
 All the other ones you closed look right, though.
 
 [Kudos to Bill for adding the neat list of pr's he closed to
 misc/4028 which made it possible to figure-out that one of
 kern/3428 and kern/3447 should be reopened.  :-]
 
 [Incidentally, the pr# 3428/3447 includes a patch, so it might
 be an easy one to fix! :-]
 
 
 --
 Outnumbered?  Maybe.  Outspoken?  Never!
 tIM...HOEk
 

From: j@uriah.heep.sax.de (J Wunsch)
To: hoek@hwcn.org
Cc: steve@FreeBSD.ORG (Steve Price), steve@hub.freebsd.org,
        freebsd-bugs@hub.freebsd.org, freebsd-gnats-submit@FreeBSD.ORG
Subject: Re: misc/4028
Date: Sun, 24 Aug 1997 00:46:48 +0200

 As Tim Vanderhoek wrote:
 
 > [Incidentally, the pr# 3428/3447 includes a patch, so it might
 > be an easy one to fix! :-]
 
 The patch looks ok.  Anyway, my local copy of st.c is contaminated
 with a number of other modifications, so it would be hard for me to
 test this here, and still cleanly extract the diff then.  Maybe
 somebody should commit it to -current (even blindly).
 
 -- 
 cheers, J"org
 
 joerg_wunsch@uriah.heep.sax.de -- http://www.sax.de/~joerg/ -- NIC: JW11-RIPE
 Never trust an operating system you don't have sources for. ;-)
>Unformatted:
