On Fri, Jan 09, 2004 at 12:36:40PM +0100, Zeremski Boris wrote:
> I can't find on Internet how to override limitation creation user names
> with dot (example firs.last). This worked in previous distribution.
>
> Any patch to shadow-utils-4.0.3-12 to allow this?
I know of no patch. None popped up with a quick google search.
Just too many dot's and dots out there.
"vipw" and "vigr" will let you modify users and groups if you need to.
Hacking shadow-utils source should be possible -- almost simple...
The check in the python code of redhat-config-users would be simple to
chop out.
As others said -- who knows what if anything will break. Adding dots
by hand does work with no instant trouble.
For now it is a very bad idea to install users that have 'meta'
characters included in their names. By meta characters I mean
characters used by regular expressions or the shell (! @ # $ % ^ & * (
) { } [ ] + - = _ | : ; ~ ` ' " , . / \ < > ? ).
The classic dislike and resistance for a user name like "fred.user" is
mostly from the "chown" command. The command "chown user:group" can
also be entered as "chown user.group". This is for compatibility with
history and interoperability with other un.x vendors. Adding dots
opens the door for confusion where a group "bob" and a user
"billy.bob" cause unexpected things to happen (errors, security
lapses).
If you hack shadow-utils you also have to hack "chown" to disallow
"user:group" == "user.group" usage to avoid the simple errors.
I can think of no good reason for pursuing this other than delivery of
mail. Aliases work fine where first.last seems to be a common thing
in the M$ world. Imap/pop may push you in this direction. In my
experience it is a simple minded idea to provide a 'friendly' name
space. The idea fails quicker than you might expect. For mail delivery
in small companies it does seem to work.
Today, adding dots into user names is in like including spaces in file
names, stupid, but possible. Same for mixed upper and lower case user
names. Consider all the mulch you can generate by fiddling with IFS.
Support for multi-byte character sets is removing much of the
downstream troubles with stuff like this. I think this is
hard stuff to do but good...
A more worthy windmill might be string handling libraries and
utilities. Null terminated strings should be a thing of the past.
--
T o m M i t c h e l l
mitch48-at-sbcglobal-dot-net
