Re: problems with latest x version in incoming
On Sun, Oct 25, 1998 at 09:47:50AM -0500, Daniel Martin wrote:
> "Adam J. Klein" <email@example.com> writes:
> > On Thu, Oct 22, 1998 at 12:35:52AM -0700, Joey Hess wrote:
> > > Adam J. Klein wrote:
> > > > This is rather strange. I took over this package relatively recently. I
> > > > hadn't really looked at the documentation.
> > >
> > > Do you have time to fully look at how suidmanager works and find out if and
> > > why the chown is necessary if suidregister doesn't run?
> > It looks like it's not neccessary. I'm not sure what the original
> > intention was, but it should be safe to change debhelper and debstd to
> > inserting only:
> > if [-x /usr/sbin/suidregister ]; then
> > suidregister -s PACKAGE FILE OWNER GROUP PERMS
> > fi
> > in the postinst. I'll fix the documentation in the next release. I'm planning
> > to rewrite suidmanager in Perl, which I hope will give me a better understanding
> > of it's exact behavior, and make it easier to add features.
> It most certainly _is_ necessary!
> As many people have pointed out, doing it this way (that is, ship the
> binary suid root in the package) as opposed to shipping the binary in
> the package not suid root and doing a chmod means that there is a
> window between the time when a package is unpacked and the time it is
> configured during which there's an suid root binary sitting out there
> that isn't being managed by suidregister.
> And this window is not small. For example, if I use dpkg -i to
> manually install some package, yet forget to install one of the
> packages it Depends: on, the package will be unpacked but not
> configured. It will remain in that state until I hunt down everything
> it Depends: on, install them, and do a dpkg --pending --configure.
Doh! That's what I get for checking on things at 1 in the morning. I'm
sorry. Thanks for pointing that out.