Re: Removing Michael Alan Dorman <email@example.com) from the project.
On Tue, Feb 05, 2019 at 10:38:50AM -0500, Michael Alan Dorman wrote:
> I did realize, though, what had been nagging me---I think you're the
> one who started the recent thread on removing "derelict" developers,
Indeed, and hopefully you will agree that I was completely tactful about it.
> or you certainly seemed to come out for it, so I'm willing to serve as
> your guinea pig. Make your case against me.
You are perhaps misconstruing several of my comments. Since you are now
active, and did close a bug recently, perhaps this can be taken as a sign
that you intend to do that for other packages. If you do intend to work on
upgrading the packages that are rather outdated, and to close the 600-,
700-, and 800-day-old bugs that you have not yet replied to, there is no
need for any further comment on the issue. I would like nothing more than
to see that happen, and to have this thread die here.
I object to taking this issue into -devel, but if you want it that way, then
so be it. I would prefer that it were not in such a public forum, however.
I feel that the clip of my message you quoted is taken quite out of context;
therefore, I include below some background and the entire message which you
21 Oct 1997: I report a bug in mailtools. Bug logs show no reply
to this 470-day-old bug to this day.
8 Nov 1997 through 23 Dec 1997: I report other bugs in your packages,
all of which are now over 400 days old. I believe that
there has not been a reply to even one.
18 Oct 1998: I report a release-critical bug in one of your pacakges.
19 Oct 1998: I send you a patch to fix that release-critical bug.
There was never a reply.
24 Oct 1998: Having received no answer to the 18 Oct mail, I NMU
libapache-mod-put, closing a release-critical bug.
To this day, I have not received a reply, and
that bug (#28135) is now sitting in "fixed" and is
over 100 days old.
17 Jan 1999: The DailyUpdate package is orphaned, in part because
your libwww-perl package is too old for many current
programs to work. I take the package and submit a bug
report against libwww-perl that is ignored.
23 Jan 1999: I politely ask your permission to ask the WNPP people
to orphan your packages after having no reply to numerous
e-mails, esp. in light of the fact that many bugs have
had no activity (not even a reply) for over 700 days.
There was never a reply.
23 Jan 1999: I NMU libwww-perl so liburi-perl may be introducted into
the distro, after numerous other developers agreed this was
the right thing to do. This was ignored as well.
5 Feb 1999: You claim that it's no problem to have 700-day-old bugs
and that I had no right to NMU your packages.
On my correspondence of 5 Feb 1999, you were upset at me for using a
"condescening tone." I pointed out that I had asked you, politely, on
numerous occasions to fix the bugs pending in your packages. You told me
that you may have just deleted my messages without reading them, and that I
ought to use the bug system. I pointed out that I have, and that you have
ignored the reported bugs for over 400 days.
You said you were "hopping mad" that I NMU'd one of your packages, even
though you made no effort to even acknowledge that the package existed since
last August (many major things had changed, and you hadn't updated the
package or responded to my bug report).
Here is my entire message. For those of you unwittingly dragged into this
thread, the first part is about mhonarc, the current package of which is
rather outdated and has some nasty bugs.
Date: Fri, 5 Feb 1999 14:04:31 -0600
From: John Goerzen <firstname.lastname@example.org>
To: Michael Alan Dorman <email@example.com>
Subject: Re: Upgrade your Multias...
References: <firstname.lastname@example.org> <19990205111952.B442@complete.org> <email@example.com> <19990205123907.A597@complete.org> <firstname.lastname@example.org>
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.95.1i
In-Reply-To: <email@example.com>; from Michael Alan Dorman on Tue, Feb 05, 2019 at 09:35:46AM -0500
On Tue, Feb 05, 2019 at 09:35:46AM -0500, Michael Alan Dorman wrote:
> If we were to not mention this change until the postinst or the
> README.Debian, we'd be too late. We may have just corrupted data for
> which there may be no other archive.
How about, then, a preinst script that checks if the person is upgrading,
and if so, from which version and to which version they are going? If they
are going between versions in which the data format changes, advise them of
this and give them the opportunity to abort the installation.
I was under the impression, though, that this only applies to lists on which
mhonarc is run in the "add" mode (obviously, it wouldn't apply to old ones
which mhonarc would not touch)
> > that they will revert to the old format in the future -- I believe
> > there is no reason to believe that with mhonarc, and you will have
> > to upgrade it sometime -- why not now?
> The variables in that file control things like umasks, data formats,
> etc. If the script is being run automatically, then we must assure
> that that data is dealt with first, properly, before we ever put the
> new version in place. Else we risk data corruption. There's no
> guarantee that if variables in that file don't get updated correctly
I'm not quite sure what "that file" here is, can you elaborate? (Not trying
to be nasty or anything, I really don't know what file it is that you're
> Do you agree this is not as trivial as it sounds? Or do you really
> have a solution that addresses the issues?
Well, the problem is that most people that use mhonarc would put the
archives some place where we would not know ahead of time. So it's not
possible to do an automatic conversion of the files; therefore, I suggest
giving them the opportunity to opt out at preinst time.
> BTW, I'll note that Jim's bug is marked "wishlist". I didn't make it
> that priority, so obviously he did---so I hope you'll understand if I
> have been taking some time to try and get it right, rather than
> rushing out because of an obvious burning need.
There are other problems, though, besides what he has been experiencing.
And there's no changelog at all (another bug!), so I'm not sure just when
the last upload was, but the current version is rather old. However, I do
understand the issues at work with this upgrade, and am certainly in favor
if getting it out correctly at the expense of a bit of extra time.
> the same time as the updated libwww-perl, so I was hesitant to
> introduce changes like this with slink and potato in such wretched
> flux during our extended freeze.
The NMU did not go into slink.
> So how do you justify the statement that it was old enough to be
> causing serious trouble for others? Where is the evidence to
> substantiate this claim?
The evidence is in the bug I reported, where I believe I said that same
thing. Several newer projects, such as DailyUpdate which T.V. orphaned
because of this very problem, require the services of both libwww-perl and
the new URI libraries; however, with the old libwww-perl, it is not possible
to do this, since the old libwww-perl and the new liburi-perl necessarily
conflict (due to using some of the same filenames).
> Now I will apologize for the fact that I don't respond *IMMEDIATELY*
> to bugs---my schedule is such that I usually have to batch them up,
I do not now, nor did I before, ask for immediate response to bug reports
unless it is a critical security concern or some such.
> and I can only get to it once a month if it's not drop-dead important.
That package had not been updated since last August, although numerous
releases had occured since then.
> > Another bug for the same package had already gone unanswered for 16
> > days, and bugs for other packages you maintain for over 700. Thus,
> "Unanswered" or "unclosed"?
Both. Most of the bugs have no answer at all from you. Many of the older
ones have only the original submission and an automated "this bug is
extremely old" message. No acknowledgement that you've even received them.
> doesn't work with package BLARF", where the original submitter has
> never responded to requests to try and reproduce the bug.
The bug logs show no attempt to contact the original submitter on all the
ones I've checked.
> Yet I don't take them out because it's impossible to know if they're
> fixed. Should I pretend I know they are?
No, but you could at least note the situation in the logs. You could try to
reproduce the problem. You could ask the original submittor to try it with
a recent version. You could ask people in -devel to try it with a recent
version. If the original submitter does not reply, or says that it cannot
be reproduced with a recent version, then you are within your rights to
close the bug.
> Also, please remember that we're in a freeze of a release, and we have
> a large number of other packages in flux (glib/gtk, for instance), so
> it is not inappropriate---although maybe conservative---to see if
> they're going to settle down before making new releases of things that
> people aren't clamoring for.
There's no reason for the freeze of slink to keep you from releasing things
into potato. Things are only made worse if bugs are left in the
distribution any longer than necessary.
> Try the bug system. That's what it's there for!
I have. You've ignored one of my bug reports for 470 days, another for 451
days, two for 406 days, libwww-perl went without comment for 16 days.
This is not the first time I've had to NMU your packages, either. And
looking at the accumulation of "Fixed" bugs, it looks like I'm not the only
one to have to do so.
I've had to twice NMU libapache-mod-put, my bug report for which you have
ignored for 108 days. That one, I reported a bug on Oct 19 and waited until
Oct 24 to fix. I am unsure at the moment if I've released the second NMU of
that; it looks like that one has not been uploaded. Again, no e-mail at all
for over 100 days.
> You reference two packages that have two bugs 43, 39 and 16 days
> old---one of which is marked wishlist and has data-loss implications
> if handled badly, to boot---and talk about how "important project
> packages are ignored?"
You have ignored bugs in sysutils, an important one.
> Why aren't you doing NMUs of dpkg, then? It's vastly more important
> than the two packages you referenced, and in much worse shape.
Because I do not have the necessary knowledge to heck dpkg's source.
> Really, reply if you wish, but I'll tell you up front that I won't
> continue to participate in this conversation if you continue with the
> tone you have established.
If you intend to trivialize ignoring bugs for hundreds of days, we can
always take the thread to -devel. I have been writing you privately because
I am hoping to get some positive response this way without causing any undue
embarassment in public.