Re: exploring debian's users and groups
On Wed, Aug 15, 2001 at 09:37:57AM -0400, Sam Hartman wrote:
> Many of the groups you state as local policy probably want to be kept
> around for compatibility with programs people might compile/install on
> their own.
Certainly, no users/groups should be unconditionally removed from
existing systems. The only problem for new systems would be if
common third party software had hard-coded dependencies on these
users/groups. Does any software explicitly rely on sync or staff?
If so, does it use these users/groups in a logical, consistent way
that we could document?
> I'd be shocked to find a Unix system without bin or adm
> users or without an adm group.
adm seems to have a useful purpose in Debian (even if not codified
in policy) and should IMO stay. bin seems to have no useful purpose
(nobody identified one). It hardly matters if people are shocked to
find something useless missing (unless people have a sentimental
attachment to bin :-) ).
> We should also consider specifications like LSB and make sure we don't
> remove users they mandate.
http://www.linuxbase.org/spec/gLSB/gLSB/usernames.html
And guess what? It is a random list of names with no useful
descriptions or rationale, and barely a hint as to how LSB-compliant
programs can use them. (I believe Ted Ts'o even said during the
recent discussion on this list that they were blindly copied from
existing systems.) This is a terrible thing for a standard, because
it encourages implementors and users of the standard to apply their
own interprentations, but that's a separate topic.
The spirit of your point stands.
Andrew
Reply to: