Re: real LSB compliance
Ethan Benson wrote:
> you would think they would know history, this was done away with long
> ago because of NFS mounted /usr.
I wish I knew the rationalle. Maybe they want lsb packages to be able to
unpack binary files owned by user bin? But that's quite silly, because
rpm should allow use of symbolic user names in the cpio (I'd hope).
They only specify 3 user id numbers: root, daemon, and bin. Very strange.
> > * LSB says that if a "operator" user exists, it should be in group root.
> > We have a group "operator" instead.
>
> what is operator even used for anyway?
No idea. That's a standard Debian really needs -- what each system user is
meant to be used for.
> this looks like more `lets
> look at redhat and make whatever we see `standard''
No, there were too few for that.
> > * LSB says that nobody's group should be called "nobody", while we have
> > "nogroup". Bah.
>
> OpenBSD has both a nobody and nogroup :shrug:
>
> if we want to bother caring about this one just add a nobody group in
> addition to the nogroup group.
Ok, the lsb package could theoretically do that.
(Of course, the sheer evilness of user nobody being in two groups called
'nobody' and 'nogroup' which should both own no files .. I guess they could be
aliases for the same group id.)
> more `lets make whatever redhat does standard!'
Of course, it's worth noting that the section on how init scripts are called
is a clear case of 'lets make whatever Debian does standard!', as is the
section on /etc/cron.{d,daily,weekly,monthly}.
--
see shy jo
Reply to: