Hi all,
On Thu, Jan 12, 2006 at 06:47:09AM +0100, Christian Perrier wrote:
>
> > > So, I hereby propose:
> >
> > > 1) close #308812 and #347584 with "Version: 3.0.20b-3
> >
> > Yes, closing 308812 since this dependency has now been added.
>
> And also close 347584 which is the same....doing so right now.
(I had already closed #347584 which I mistakenly cloned, but that does not
really matter anyway)
Both #308812 and #347585 still affect sarge, which is currently our
official release. Under normal circumstances I don't usually care about
bugs being closed if they are in stable but fixed in unstable (and
possibly testing). However, both of these simply make the software plain
unusable.
I'd suggest reopening #308812 and then tagging sarge and fixed, so it
stays documented in BTS. And perhaps rebuild for sarge with necessary
dependency and upload to proposed-updates, though I'm not sure without
checking what qualifications are these days for non-security issues to
still get through to stable point releases.
> > > 2) lower #347585 to important
> >
> > > I don't do it right now because I really want to understand why you
> > > cloned the bug.
> >
> > I believe this one should still be left open, and at the present severity.
>
> Well, my rationale is that #347585 is not about a missing dependency
> but about a missing directory, which only affects those people who use
> winbind without samba.
Which is a perfectly fine use case.
Winbind is about authenticating (anything) to AD. As I wanted pppd on a
VPN gateway to do so, I had no need for samba on this machine, at all. If
I want squid to use NTLM auth, I have no need for samba. If I want cyrus
to use NTLM auth, ... You can guess it.
Heck, even on a workstation where I would want to use NTLM auth I probably
would not want to install samba or possibly even a samba client (be it
smbmount or smbclient).
Additionally, while the ntlm_auth binary complains when it is affected by
#308812, winbind just dies with no explanation whatsoever because of
#347585 - even when using debuglevel up to 2048. It took an invocation
with strace to find it out why.
> The solution to #347585 is not another dependency but rather
> /var/run/samba included in winbind so one may argue that, seen this
> way, the bug is something that "has a major effect on the
> usability of a package, without rendering it completely unusable to
> everyone".....:-)
I'd argue it makes winbind completely unusable in the use case which is
according to the description of the package the reason why it exists.
> Anyway, this is arguing: fixing this give us an opportunity to upload
> a new version which I was already considering given that the former
> one has reached testing.
Ack :)
However, as explained above, just closing (by whatever method) the bugs is
IMHO inappropriate given that they both affect stable.
Regards,
Filip
--
/* Amuse the user. */
\|/ ____ \|/
"@'/ ,. \`@"
/_| \__/ |_\
\__U_/
-- /usr/src/linux-2.4.2/arch/sparc/kernel/traps.c::die_if_kernel()
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
