On Wed, Dec 28, 2005 at 12:30:53PM +0100, Javier Fernández-Sanguino Peña wrote:
> On Wed, Dec 28, 2005 at 03:12:44AM -0800, Steve Langasek wrote:

> > > Since there is no libssl097-dev any longer I guess I'll have to recompile 
> > > all
> > > packages.

> > It should actually be possible to fix this with binNMUs on the autobuilders,
> > I think.  I'll go ahead and queue those now.

> Please don't. The libssl 0.9.8 does *not* work when using Nessus, I've just
> recompiled all packages (libnasl, nessus-plugins and nessus-core) to try to
> get it working and I still get this:

Already done, though; as I said, it doesn't make things any *worse*, and
this is an RC bug in libssl0.9.8 that needs to be fixed.  Having libnasl
stay linked against libssl0.9.7, and then accidentally get broken in a
security reupload, wouldn't be good either, so we might as well have
binaries in the archive that correspond to the current sources.

On Wed, Dec 28, 2005 at 12:47:48PM +0100, Javier Fernández-Sanguino Peña wrote:
> On Wed, Dec 28, 2005 at 02:54:17AM -0800, Steve Langasek wrote:

> > > * nessusd 2.2.5-3, the server, is linked against both 0.9.7 and
> > > 0.9.8

> > Ok, I don't see this either:

> > $ ldd /tmp/nessus/usr/sbin/nessusd|grep ssl
> >         libssl.so.0.9.8 => not found
> > $

> Funny, it seems that ldd output varies _if_ you have this:

Right... the problem with ldd is that it recurses library dependencies, so
it doesn't really tell you where the problem lies. :)

> So, for archs that have compiled libnasl2 against libssl.so.0.9.8 you will
> not "see" nessusd linking against both. For archs that have compiled libnasl
> aginast libssl.so.0.9.7 you will see that. Tthose archs include i386 at
> least, since the packages for i386 were compiled in August by me. Which was
> previous to the switch of 0.9.7 to 0.9.8 in libssl-dev (in October).

It was actually the case on all architectures, fwiw.

> > To me, this bug looks like it's just an instance
> > of #338006.

> Indeed, it looks like this might be the end issue. Is it a good idea to force
> everyone to use a buggy library? Wouldn't it make sense to provide a
> libssl097-dev to prevent breakage for those packages that get bitten by this
> bug?

As mentioned, the bug in libssl0.9.8 *is* RC; and I don't think we're going
to be reverting all of these packages to remove libssl0.9.8 from etch; so I
believe it's better to focus on fixing openssl instead of trying to work
around it.

In the meantime, I guess I would have to recommend that users who need
nessus use the version from stable.

-- 
Steve Langasek                   Give me a lever long enough and a Free OS
Debian Developer                   to set it on, and I can move the world.
[EMAIL PROTECTED]                                   http://www.debian.org/

Attachment: signature.asc
Description: Digital signature

Reply via email to