Instead it uses the
dpkg-trigger command.
So no objection here to the change in default behavior.
--
Thomas Hood
Resolvconf maintainers
--
To UNSUBSCRIBE, email to debian-dpkg-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/4dea129c.6060...@gmail.com
On 3 September 2014 10:28, Johannes Schauer wrote:
> Can we somehow fix that?
>
> CC-ing Thomas Hood as he introduced that change into the spec.
Please note that at that time (Octobler 2013) I was trying to clarify
and expand the explanation in order to reflect what I thought was
inte
On Thu, 2003-07-24 at 13:46, Stephen Frost wrote:
> I see this as totally bogus. Either the conffile is shared or it isn't.
> If it's shared then the packages involved know this
Package foo which eliminates /etc/foo.conf doesn't "know"
that there is not some other package, bar, which Depends
on f
On Thu, 2003-07-24 at 14:53, Stephen Frost wrote:
> * Thomas Hood ([EMAIL PROTECTED]) wrote:
> > Package foo which eliminates /etc/foo.conf doesn't "know"
> > that there is not some other package, bar, which Depends
> > on foo and uses /etc/foo.conf . That
On Fri, 2003-07-25 at 07:59, Manoj Srivastava wrote:
> Umm. apt allows you to determine reverse depends. From there
> there is an easy hop to sending email to ask the develoeprsa in
> question; or to exaimine a package to look at its conffiles.
This doesn't solve the problem of the depen
ot; after
"start-stop-daemon --stop --pidfile $PIDFILE".
--
Thomas Hood <[EMAIL PROTECTED]>
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
This bugs still exists.
_
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com
This RC bug was reassigned to dpkg several days ago, but
there has been no comment from the maintainers. What's
happening?
signature.asc
Description: This is a digitally signed message part
This RC bug was reported several days ago but the maintainer
has not responded. What's happening?
_
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject o
Package: dpkg
Version: 1.9.20
Severity: serious
Justification: Policy 2.2
Package is required and has a Pre-Depends on libstdc++3 (>= 1:3.0.3-1) which is
standard on hppa
-- System Information
Debian Release: 3.0
Architecture: i386
Kernel: Linux thanatos 2.4.19-pre7-ac2 #2 Sat Apr 20 16:51:23 ED
severity 160761 normal
merge 160761 3410
thanks
I see that report #3410, which already reported this bug,
is 6 years and 103 days old.
Cheers
--
Thomas
Package: dpkg
Version: 1.10.4
Severity: wishlist
It would be useful if dpkg -l would enlarge the "Name" field
sufficiently to display the longest package name in the list.
Perhaps this behavior should be selected by an option.
With the current behavior, I sometimes have to enlarge my
X terminal
Orn E. Hansen wrote: [I have edited it for grammar]
> Replacing a package with another that provides the same package.
> causes conflicts with programs that depend on it... an example:
> I have quota and e2fsprogs installed. They are marked essential.
> I try to replace e2fsprogs with e2compr, a
Housekeeping ...
> Date: Mon, 15 Feb 1999 00:41:59 + (GMT)
>
> dpkg --forget-old-unavail only removes unavailable packages
> from the status file which have Status: purge ok not-installed,
> but not those which have Status: hold ok not-installed,
This doesn't look like a bug to me. Close?
-
severity 41554 minor
tags 41554 patch
thanks
The complaint here was that when the pidfile method
is used to check whether an instance of a daemon is
running, it is the existence of the pidfile and not the
existence of the process named in the pidfile that
is checked.
This is reasonable behavior.
Housekeeping ...
Report 60028 was tagged "pending" on 19 Nov 2001.
Can it now be closed?
--
Thomas
This report was tagged "pending" on 19 Nov 2001.
Can it now be closed?
--
Thomas
generated on the second pass of the retry sequence.
The return value is 0.
dpkg Version: 1.10.4
--
Thomas Hood
t system knows about?
This file isn't part of the GNU info system, SFAICT.
The man page needs to be corrected.
--
Thomas Hood
Housekeeping ...
If I understand correctly, there is no problem here (#156142).
Close?
reopen 33394
severity 33394 wishlist
thanks
OK, let's keep this open as a wishlist item.
The wish here is that --forget-old-unavail make dpkg
forget about packages that are "hold ok not-installed"
as well as about those that are "purge ok not-installed",
since some people still have records of ne
severity 109691 normal
tags 109691 moreinfo
thanks
> I've been trying to get a sendmail-tls package setup so that one could
> seemlessly change from sendmail to sendmail-tls (and back again).
> The conffiles names and contents are *identicle* in the two packages,
> and they each 'Replaces:' the o
.postrm
#!/bin/sh
set -e
if [ -x /usr/bin/update-menus ]; then
update-menus
fi
if [ -x /usr/sbin/update-mime ]; then
update-mime
fi
if [ "$1" = "purge" ]; then
rm -f /etc/lynx.cfg
fi
--
Thomas Hood
etd.conf, even though this now
belongs to ftpd-ssl.
This *may* explain the original problem that was reported.
There may also be a bug in dpkg, but let's follow that
one at #109691.
Apologies if I am somehow confused here.
--
Thomas Hood
update-rc.d -f exim remove >/dev/null
# Remove from inetd.conf
update-inetd --remove exim
;;
upgrade|failed-upgrade|abort-install|abort-upgrade|disappear)
;;
esac
--
Thomas Hood
mes purges conffiles that have been
taken over by replacing packages. The thing is that every
case of this phenomenon I have investigated can be attributed
to a misbehaving postrm. Therefore, I am inclined to think that
109691 can be closed soon unless someone can prove that there
really is a fault
es should
be left to dpkg to delete on purge.
Can the originator of the report (Bill Gribble) reproduce the
bug? If not, then I suggest that this report be closed,
as it probably concerns an obsolete package.
--
Thomas Hood
(This is information for #47267, but I am cc:ing #163183
since I make reference to it below.)
On Thu, 2002-10-03 at 15:23, Ben Collins wrote:
> On Thu, Oct 03, 2002 at 02:19:55PM +0200, Thomas Hood wrote:
> > I have a strong suspicion that the fault here lies with the
> > p
I have tested the hypothesis that dpkg will delete a
conffile that has been taken over by a Conflicts:ing
package. I used two packages that both declare the
same conffile, the second package Conflicts:ing with
the first package.
1. Install #1 (which installs the conffile)
2. Install #2 (wh
On Thu, 2002-10-03 at 16:48, Ben Collins wrote:
> > I just tested your hypothesis on two packages that
> > share a conffile, the second Conflicts:ing with the
> > first (as login Conflicts: with pam-apps):
> > 1. Install #1 (which installs the conffile)
> > 2. Install #2 (which removes #1 a
information in his original post.
Bill Gribble (or anyone): Can you reproduce this misbehavior?
--
Thomas Hood
Adam Heath wrote:
> I don't care what you have seen.
Regardless, I went ahead and conducted more experiments using
a couple of packages I cooked up to test dpkg.
I found that when the two packages had the same conffile,
and the second was declared as either Conflicts:ing with the first
or Replace
I just tried to reproduce #53380 and was unable to.
The current version of ntpdate ( 1:4.1.0-8) preserves user
changes to /etc/init.d/ntpdate.
Can this report be closed, or do you have more info?
--
Thomas
view? If porting to date -R is planned,
this report can be set to severity wishlist. Otherwise this
report can be closed.
--
Thomas Hood
problem
may be related to #109691 and the reports merged with it.
(I am cc:ing #109691 on this.)
Anyone have any more info?
--
Thomas Hood
he preinst
and the postinst are implicated.
I have just looked at the maintainer scripts of netbase
version 4.07 and there are no references in there to
/etc/inetd.conf. However similar code has appeared in
netkit-inetd. So I'm reassigning this report to that
package.
--
Thomas Hood
go through this draft carefully. (It's in sgml, but it
is easy to read between the tags.) I have included some
questions in BLOCK LETTERS to which I would appreciate answers.
Thanks
// Thomas Hood
%aptent;
]>
&
field could be invented that declared
a stronger conflict that Conflicts: does -- namely, a
conflict that requires purging of the conflicting package
before installation of the replacing one.
--
Thomas Hood
-- System Information:
Debian Release: 3.0
Architecture: i386
Kernel: Linux thanatos 2.4.20-
Housekeeping ...
As the maintainers wish to keep dpkg portable, they have
said that they do not want to make dpkg depend on textutils
(which was the wish here).
So, can this report be closed now?
--
Thomas
> Mention of apt would be appropriate now.
Indeed.
> Also perhaps installing a link for a manpage
> of dpkg.conf to view the dpkg man page would be great.
I'm not sure what you mean. There exists a man page
dpkg.cfg(5) but what sort of "link" do you have in mind?
dpkg.cfg(5) and apt(8) should
This one is weird. Diversions aside, there is no way
vim-gtk should disappear on installation of vim.
vim-gtk depends on vim, and none of its files overlap vim's,
except for /usr/bin/vim, which is handled by the diversion.
Can you reproduce this? Is it possible that you installed
a *newer* versi
severity 82407 wishlist
retitle 82407 Please don't leave behind unpacked files on install abort
thanks
> so dpkg won't install erlang-doc-html when erlang is installed
> but it leaves unpacked files from erlang-doc-html on disk:
There is a bug in the erlang and/or erlang-doc-html packages,
not in
Is this a bug or a feature?
> If an error occurs with dpkg --unpack (or in the
> equivalent stage of dpkg --install), then all details
> about the package filenames are removed from the
> available database.
If this isn't a bug then this report should be downgraded or closed.
--
Thomas Hood
-
dpkg: considering removing kbd in favour of kbd-compat ...
dpkg: yes, will remove kbd in favour of kbd-compat.
[...]
dpkg: dependency problems prevent configuration of kbd-compat:
kbd-compat depends on console-tools (>= 1997.10.28-1); however:
Package consol
"dpkg-divert --remove" always moves the diverted file
back to the original name. Is this a bug, or should
the behavior simply be documented?
--
Thomas
to ask about replacing diverted conffiles as well as normal
> conffiles, maybe stating that the file has been diverted.
Sounds reasonable.
--
Thomas Hood
that --remove does not do this unless the
--rename option is also specified.
--
Thomas Hood
Ian Jackson wrote:
> I'll look out for something like this, and leave the bug open as a
> record in case anything recurs. If there are no further reports I'll
> probably close it in a year or so ...
That was over three years ago.
Can this report be closed now?
--
Thomas Hood
reassign 61663 linuxlogo
severity 61663 serious
thanks
Wichert Akkerman wrote:
> Diverting conffiles is not supported and will break things.
So this isn't a fault in dpkg.
linuxlogo's scripts futz with a conffile. That appears to
contravene policy (see 11.7.3) so I am reassigning this to
linuxl
Hi.
Was your diversions file corrupt?
Have you fixed the problem? (Report was dated Nov 1998.)
Did it occur again?
Can this report now be closed?
--
Thomas Hood
r/cache/apt/archives/inewsinn_1.7.2-11_i386.deb
The behavior described in 46716 looks like some sort
of external problem: perhaps the filesystem
was mounted read-only (or something along those lines)?
Did you ever figure out what was happening?
--
Thomas Hood
severity 47839 wishlist
thanks
The wish is for ability to regenerate corrupted .list files.
Reported:
> $ dpkg --compare-versions 200101102349 lt 200103301546; echo $?
> 1
> This isn't correct.
Now:
$ dpkg --compare-versions 200101102349 lt 200103301546; echo $?
0
I take it the patch went in and the problem was fixed.
Close?
--
Thomas
Provides: gs-suitable-for-magicfilter
magicfilter
Suggests: gs-suitable-for-magicfilter
You get the idea.
--
Thomas Hood
Wichert wrote:
> While doing "dpkg -iGOEB *deb" in a directory I noticed this:
>
> Preparing to replace ssh 1.2.26-1 (using ssh_1.2.26-1.2.deb) ...
> Stopping Secure Shell server: sshd.
> Unpacking replacement ssh ...
> Preparing to replace ssh 1.2.26-1.2 (using ssh_1.2.26-1.deb) ...
> Unpacking re
tags 33994 moreinfo
thanks
I just tried this and "fg" did not cause apt to
clear the screen. Which is not to say that this
can never happen. Under what circumstances does
this happen?
--
Thomas
-pre8-ac1
Status: install ok installed
Priority: extra
Section: utils
Installed-Size: 100
Maintainer: Thomas Hood <[EMAIL PROTECTED]>
Version: 3.7-1+10.00.Custom
Replaces: tpctl-modules-2.4.20-pre8-ac1
Provides: thinkpad-modules
Depends: thinkpad-base, modutils
Suggests: devfsd
Conflicts:
154633: install-info: --help should use stdout
103563: install-info: --quiet hides error messages
Informational messages should be on stdout, error messages
on stderr, and --quiet should reduce informational messages,
not error messages.
gradall:/home/adam# grep-status -F Package jboss-server -n -s Conffiles|
awk
'{ print $2 " " $1 }' | sort -u | wc -l
86
gradall:/home/adam# grep-status -F Package jboss-server -n -s Conffiles|
awk
'{ print $2 " " $1 }' | sort | wc -l
170
This bug can't be addressed unless more information is
provided.
The patch you provided doesn't seem to apply to
dpkg-shlibdeps any more. Has the bug been fixed?
(#20794)
--
Thomas Hood
.d/S20logoutd
/etc/rc5.d/S20logoutd
/etc/rc6.d/K20logoutd
-
More info will be needed in order to reproduce this bug.
My guess is that one of dpkg's status files was corrupt
when this happened.
--
Thomas Hood
Being able to set the PATH for dpkg might be considered
a feature, not a bug.
--
Thomas Hood
dpkg version 1.10.4
Yep, he's right. changelog.gz still contains entries
from the year 2001 after entries from the year 1994,
beginning with line 6263.
--
Thomas Hood
On Wed, 2002-10-09 at 20:17, Paul Vojta wrote:
> > Being able to set the PATH for dpkg might be considered
> > a feature, not a bug.
>
> Only if it's done intentionally.
> E.g.: bash% DPKG_PATH=... dpkg -i foobar.deb
dpkg can't know how the path was set.
Still, I think you are right that the p
Hi Russell.
I don't think dpkg is in the wrong here. It appears that
pcmcia-modules-2.4.0-test9 requires the same release level
as kernel-image-2.4.0-test9. Installation can only
be done one package at a time, but whichever of the two
we install first will produce a version mismatch with the
oth
tags 107098 patch
tags 107098 moreinfo
thanks
textutils has always included the file
/usr/share/info/textutils.info.gz, not the file
/usr/share/info/textutils.info .
Nevertheless, if install-info is asked to install the
latter, it ought to check for the former. I can't see
why it didn't in this
severity 48895 wishlist
thanks
The wish, I take it, is for a way to remove "dangling"
alternatives without being left in manual update mode.
Does the submitter have any new information about this issue?
--
Thomas Hood
severity 98822 minor
thanks
I take it there is no misbehavior here; just an ugly
"cannot stat" message.
#158753 concerns "dpkg-query -l" and "dpkg-query -l *".
#137442 concerns "dpkg --get-selections" and "dpkg --get-selections *".
Presumably, these commands should behave in analogous ways.
It seems that dpkg-query is being changed so that "dpkg-query -l"
will produce the same listing as "dpkg-quer
On Tue, 2002-10-15 at 23:53, Adrian Bridgett wrote:
> > Is that bug (#108196) reproducible?
>
> Yep. It is caused if you edit debian/conffiles but forget to have a trailing
> newline (it just so happened I didn't press enter whilst creating it and the
> editor didn't automatically add one) - i.e i
Hi, and thanks for the additional info.
dpkg is actually behaving correctly here.
In order to install xlibmesag3 dpkg has to remove
mesag3 first. However, removing mesag3 would break
the dependency on libgl1. So, dpkg cannot proceed.
Changing from one provider of libgl1 to another is
different
Here's my latest draft of the apt_preferences(5) man page.
Please read it over and report any errors!// Thomas
%aptent;
]>
&apt-docinfo;
apt_preferences
5
apt_preferences
Preference control file for APT
Description
The APT preferences file /etc/apt/
[EMAIL PROTECTED]:~$ md5sum .bashrc
f1075ff35a46c75de41b84ac3039a8de .bashrc
[EMAIL PROTECTED]:~$ md5sum < .bashrc
f1075ff35a46c75de41b84ac3039a8de -
Please follow up at #164591.
--
Thomas Hood
Package: dpkg
Version: 1.10.4
Severity: normal
I get:
dpkg: autoconf2.13: dependency problems, but removing anyway as you request:
autoconf depends on autoconf2.13 (>= 2.13-41).
Removing autoconf2.13 ...
syntax error at /usr/sbin/install-info line 128, near "else if"
syntax error at /usr/sbin/ins
Package: dpkg
Version: 1.10.9
Severity: minor
Excerpt from the man page:
=
generic name
A name, like /usr/bin/editor, which refers, via the
alternatives system, to one of a number of files of
I have been able to reproduce this bug on my own machine.
In the following procedure, step 4 incorrectly removes
conffiles that have been taken over by bind9.
1. Install bind
2. Remove bind
3. Install bind9
4. Purge bind
A full console log is attached.
[EMAIL PROTECTED]:/var/cache/apt/archives$
Herbert Xu <[EMAIL PROTECTED]> wrote (to debian-devel):
> Unfortunately dpkg does not handle the case where a conffile
> ceases to exist in a later version of a package. The conffile
> will be left on the system even after purging.
Agreed, dpkg should do something reasonable when a package
is rep
ame time, one of these
packages must be defined as owner of the configuration file, i.e.,
it will be the package which handles that file as a configuration
file. Other packages that use the configuration file must depend
on the owning package if they require the configuration file to
operate.
--
Thomas Hood
On Tue, 2003-07-22 at 06:30, Branden Robinson wrote:
> On Tue, Jul 22, 2003 at 07:21:04AM +1000, Herbert Xu wrote:
> > What about this?
> >
> > Treat the disappearence of a conffile the same way as you treat any
> > other changes. Prompt the user as usual, if they answer yes, then
> > rename the
On Thu, 2003-07-24 at 08:47, Andreas Metzler wrote:
> It really sucks to handle this if you want/need to get rid of it (if
> it is unmodified) not only on purge but on upgrades. - You'll need
>
> if [ "$1" = "configure" ] && \
> dpkg --compare-versions "$2" le-nl "1.2.3" && \
> [
I think we have heard all sides of this now (in debian-devel)
I conclude that dpkg should deal with an obsolete conffile in
the following way.
If the conffile has not been locally modified, delete it.
If the conffile has been locally modified, ask the user for
permission to eliminate the file; if
ve sgen slink salt] ...
> --install genname symlink altern priority [--slave sgen slink salt] ...
--
Thomas Hood <[EMAIL PROTECTED]>
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Package: dpkg-doc
Version: 1.10.26
Severity: minor
"command -v" is not POSIX. However, the postinst and prerm scripts use
it and have #!/bin/sh at the top, so they violate policy 10.4.
Possible solutions:
* Using #!/bin/bash eliminates the 10.4 violation. However, there
remains the additiona
theirs?
Is this purely because dpkg is the more senior program, or is there some
other hindrance to deleting the space before "..." in dpkg's output?
--
Thomas Hood <[EMAIL PROTECTED]>
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Package: dpkg-dev
Version: 1.10.15
Severity: critical
Tags: sid
Justification: breaks unrelated software
I am slapping an RC bug on this just so that this issue will be
considered before 1.10.15 is allowed to slide in to sarge.
Version 1.10.10 built a thinkpad-modules-the.kern.ver package fine.
V
belong to other packages.
*/
Someone please help me with this so that we can close these nine
merged severity-important bug reports, the oldest of which has been
open for almost seven years!
--
Thomas Hood <[EMAIL PROTECTED]>
Scott James Remnant <[EMAIL PROTECTED]> wrote:
> I think I've fixed it
I think you forgot to give credit to those who helped you find the bug.
--
Thomas Hood <[EMAIL PROTECTED]>
Package: dpkg
Version: 1.10.21
Severity: normal
$ dpkg --compare-versions 1.1 lt 11 && echo yes
yes
$ dpkg --compare-versions 1.a lt 1a && echo yes
$ dpkg --compare-versions 1a lt 1.a && echo yes
yes
Why is version 1a less than 1.a?
-- System Information:
Debian Release: testing/unstable
APT p
Package: dpkg
Version: 1.10.21
Severity: wishlist
If you try to start a daemon using start-stop-daemon and the daemon
is already running then start-stop-daemon returns 1.
If the daemon is not already running and start-stop-daemon starts
an instance of it, and the instance exits with status 1, the
at "start-stop-daemon --stop"
uses the --retry option?
--
Thomas Hood
On Mon, 2004-08-30 at 07:21, GOTO Masanori wrote:
> We're talking about libc6.postinst start-stop code (line:264). We
> don't execute start-stop-daemon directly. So I think --retry option
> issue is out of scope.
One should be able to call an initscript with the "restart"
argument in order to r
92 matches
Mail list logo
