Steven Chamberlain | 18 Mar 01:05 2012

Bug#659279: netmrg: FTBFS(kfreebsd): error in kernel header


> A pair of #includes are missing.  machine/param.h is required by
> in_pcb.h and sys/socket.h is needed by if_arp.h.  See attached patch
> (untested).

I just tried building this (on kfreebsd-i386) with the new
kfreebsd-kernel-headers (0.78).  Somehow this has fixed *both* issues so
the patch/includes are no longer needed.

Christoph, please kindly schedule a rebuild of this once the new k-k-h
are ready on the buildd's.


Steven Chamberlain
steven <at>

Russ Allbery | 18 Mar 01:11 2012

Bug#663918: debian-policy: forbid deprecated relations (<, >)

Charles Plessy <plessy <at>> writes:
> Le Wed, Mar 14, 2012 at 12:28:10AM +0100, Jakub Wilk a écrit :

>> There are only 12 packages in the archive still using this confusing
>> syntax:

> One less (r-cran-psy).

> I see that rp-pppoe has not been uploaded for a long time.  It might be
> better to check that rebuilding will not introduce unforseen bugs,
> before making the current version unfit for a release.

I confirmed that rp-pppoe still builds, although with a lot of warnings
due to the use of these deprecated relations and a deprecated version of
debhelper, among other things.  I can't test it since I don't use PPP.

However, I'm comfortable going forward anyway for two reasons: one, we
want packages to be rebuildable and packages that aren't rebuildable are
already RC-buggy even if we haven't detected that problem; and two, making
it a Policy must does not automatically make it RC.  That's still at the
discretion of the release managers, and I suspect they'd be willing to
give packages a pass for something relatively minor like this.

So, I'm seconding the original change and, that making three seconds
counting the proposer, applying the following change for the next release,
which also rewords the section minorly for clarity.

diff --git a/policy.sgml b/policy.sgml
index 4fd60d9..52dbb26 100644
(Continue reading)

Russ Allbery | 18 Mar 01:12 2012

Bug#663930: debian-policy: New virtual package for naming spec compliant icon themes

Fernando Tarlá Cardoso Lemos <fernandotcl <at>> writes:

> Discussion is underway in debian-devel about proposing the creation of a
> virtual package that represents an icon theme package that follows the
> naming spec for icon themes[1].

> I'll hopefully be able to update this bug report with the outcome of the
> discussion, as requested in [2].

> [1]:
> [2]:

Thanks!  This certainly looks like a good idea to me.


Russ Allbery (rra <at>               <>

Christoph Anton Mitterer | 18 Mar 01:13 2012

Bug#664209: apt_preferences: package of higher prio not choosen?!

Hi Daniel.

On Sat, 2012-03-17 at 12:29 +0800, Daniel Hartwig wrote:
> Like David says, please always modify the version number.
Well I did this, or actually my the whole need for this has vanished in
the meantime ;-)

Nevertheless,... I still think that documentation is ambiguous/incorrect
here... or even more, that handling is incorrect: reinstalls should not
be considered as downgrades...
And so far, no one could show me why I'm wrong with respect to this.

So I guess best would be to reopen this bug and discuss whether a)
behaviour could/should be changed or b) the apt_preferences(5) manpage
could be extended to note that reinstall == downgrade, e.g. from
>·   Never downgrade unless the priority of an available version exceeds
>    1000. ("Downgrading" is installing a less recent version of a
>    package in place of a more recent version.

>·   Never downgrade (OR REINSTALL) unless the priority of an available 
>    version exceeds 1000. ("Downgrading" is installing a less recent OR
>    THE SAME version of a package in place of ANOTHER version.

(There may be other places in the manpage, where the same information
should be added.)

The above is however probably not enough, as David says,.. the order of
sources.list entries is important... so reinstalling is done if ordered
first but not the other way round.
(Continue reading)

Stephane Popinet | 18 Mar 01:25 2012

Bug#652258: Possible removal of gerris

Hi Neil et al,

>> gts-config is obsolete. You should use `pkg-config --cflags gts`
>> and/or `pkg-config --libs gts`
> That requires changes to but gerris fails with autoreconf:
> doc/examples/ `pkglibdir' is not a legitimate directory
> for `DATA' src/ `pkglibdir' is not a legitimate
> directory for `DATA' autoreconf: automake failed with exit status: 1

I am the lead developer of Gerris (and GTS). I am about to release the
attached patches (in darcs format sorry) for the latest snapshot
release of Gerris which should fix both set of issues (i.e. obsolete
gts-config and failures with latest automake).


Attachment (patches): application/octet-stream, 8 KiB
Steven Chamberlain | 18 Mar 01:41 2012

Bug#664470: proxytunnel: FTBFS on kfreebsd-* and hurd

Package: src:proxytunnel
Version: 1.9.0-3
Tags: patch
User: debian-bsd <at>
Usertags: kfreebsd
X-Debbugs-Cc: debian-bsd <at>


Something weird in the Makefile causes this error on kfreebsd and hurd

-c -o proxytunnel.o proxytunnel.c
> In file included from /usr/include/i386-kfreebsd-gnu/sys/socket.h:40:0,
>                  from /usr/include/netinet/in.h:25,
>                  from proxytunnel.c:28:
> /usr/include/i386-kfreebsd-gnu/bits/socket.h:345:5: error: expected identifier before numeric constant
> proxytunnel.c: In function ‘do_daemon’:
> proxytunnel.c:190:36: error: ‘SO_REUSEPORT’ undeclared (first use in this function)
> proxytunnel.c:190:36: note: each undeclared identifier is reported only once for each function it
appears in
> make: *** [proxytunnel.o] Error 1

By defining SO_REUSEPORT in the compile options, the sys/socket.h header
is unable to properly define it.

I think that hack was originally meant to help with Cygwin builds.  But
I've tried removing it (please see attached patch), and that fixes the
build on kfreebsd-i386 and still builds fine on (Linux) amd64 and so
(Continue reading)

Tshepang Lekhonkhobe | 18 Mar 01:43 2012

Bug#664231: wajig: uses "su -c", even when the user is already root

On Sat, Mar 17, 2012 at 13:44, Francesco Poli <invernomuto <at>> wrote:
> On Sat, 17 Mar 2012 13:11:16 +0200 Tshepang Lekhonkhobe wrote:
>> Interestingly I had already unwittingly
>> fixed this bug by getting rid of the --quiet option (version is in
>> Google code hg repo). I will close this bug once that version is
>> uploaded to Debian, which should be within a month.
> Mmmmh, it does *not* seem to me that the code currently in the Mercurial
> repository fixes the bug: the indentation of
>  command = setroot + " -c '" + command + "'"
> still looks wrong to me.
> I am attaching a new patch, this time to be applied to the version of
> currently in the Mercurial repository, to this message.
> It is my understanding that this new patch should fix the issue.

Thanks for the gentle interaction. Forgive me for my careless patch
review and thanks for your careful attention to detail. The fix is now

Norbert Preining | 18 Mar 01:48 2012

Bug#593448: scalable-cyrfonts and font naming

Hi Karl, hi Anton,

(I forgot to include the original bug report, now included)

On Sa, 17 Mär 2012, Karl Berry wrote:
> Hi Norbert,
> 	Free Times, Free Helvetian, Free Helvetian Condensed, Free Courier, 
> 	Free Avant Garde, Free Paladin, Free Schoolbook, Free Bookman,
> 	Free Chancery, Teams and OldSlavic.
> Those names seem awfully generic, and there are some trademarks in
> there ... anyway ...

I agree.

>     Now that creates a clash that is sub-optimal.
> Not just suboptimal.  Completely unacceptable.  Conflicts would make
> both fonts sets unusable, in practice.

Also here I agree.

>             fpl:	TL package fpl: SC and OsF fonts for URW Palladio L
>                                     shipping the fonts
>                     TL package psnfss: shipping fpl maps
>             fnc:	TL package fouriernc: 
> I see no magic bullet here.  We can't create a conflict.  So either the
> fpl and fnc authors change their names, the cyrfonts authors change
(Continue reading)

Iustin Pop | 18 Mar 01:51 2012

Bug#664471: false positive for hyphen-used-as-minus-sign

Package: lintian
Version: 2.5.6
Severity: minor


It seems to me that lintian falsely reports hyphen-used-as-minus-sign
for the following line in the file:

Use \f[C]--\ --help\f[] for a list.

If the space ('\ ') is removed, the warning is no longer
generated. I'm not familiar with the groff source language, but at
least viewing the page through man shows regular minus signs, and not

-- System Information:
Debian Release: wheezy/sid
  APT prefers unstable
  APT policy: (500, 'unstable'), (500, 'testing'), (1, 'experimental')
Architecture: amd64 (x86_64)

Kernel: Linux 3.2.5-ruru0 (SMP w/4 CPU cores)
Locale: LANG=ro_RO.UTF-8, LC_CTYPE=ro_RO.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

Versions of packages lintian depends on:
ii  binutils                       2.22-6
ii  bzip2                          1.0.6-1
ii  diffstat                       1.55-2
(Continue reading)

Jonathan Nieder | 18 Mar 01:50 2012

Bug#601727: no sound laptop speakers


Bünyamin wrote:


That's not an Amilo Pi.  Please file a new bug.

Thanks and sorry for the trouble,