Adduser devel - May 2005 - Bug#298883: there should be a way to override PASS_MAX_DAYS in useradd

Hi!

I just forgot about a "secret" -O option in
useradd. :)
You can use it to override whatever values from
login.defs

For example, regarding bug #298883, use of:
> `useradd -O PASS_MAX_DAYS=99999 usr298883`
will solve your problem (other solution, that you have
currently implemented, involves `chage`).

Thus closing the bug.

The problem with -O option is a Tomasz''s desire to
rename it to -K for sake of unfication with Solaris:

On Thu, Mar 24, 2005 at 11:31:45PM +0100, Tomasz Kłoczko
wrote:
> I see some good things in shadow like tools in Solaris and some things I 
> want to integrate on shadow.
...
> All above variables can be used by "useradd -K
<key>=<value>".
...
> Current shadow useradd have undocumented -O option. Fragment from
> useradd source code:
> 
>                 case ''O'':
>                         /*
>                          * override login.defs defaults (-O name=value)
>                          * example: -O UID_MIN=100 -O UID_MAX=499
>                          * note: -O UID_MIN=10,UID_MAX=499 doesn''t
work yet
>                          */
>                         cp = strchr (optarg, ''='');
>                         if (!cp) {
>                                 fprintf (stderr,
>                                          _("%s: -O requires
NAME=VALUE\n"),
>                                          Prog);
>                                 exit (E_BAD_ARG);
>                         }
>                         /* terminate name, point to value */
>                         *cp++ = ''\0'';
>                         if (putdef_str (optarg, cp) < 0)
>                                 exit (E_BAD_ARG);
>                         break;
> 
> IMO it will be good kill -O an move this to -K.
P.S. now you know the alternatives, at least. ;)

-- 
WBR,
xrgtn

reopen #304934
thanks

On Mon, May 02, 2005 at 02:30:08AM +0300, Alexander Gattin
wrote:
> I just forgot about a "secret" -O option in
> useradd. :)
> You can use it to override whatever values from
> login.defs
> 
> For example, regarding bug #298883, use of:
> > `useradd -O PASS_MAX_DAYS=99999 usr298883`
> will solve your problem (other solution, that you have
> currently implemented, involves `chage`).
> 
> Thus closing the bug.
Reopening. Please document this option. I cannot use a secret,
undocumented option.
> The problem with -O option is a Tomasz''s desire to
> rename it to -K for sake of unfication with Solaris:
So we can/need to wait until it is renamed, then document it, and then
close the bug.

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don''t trust Computers. They | Mailadresse
im Header
Mannheim, Germany  |  lose things."    Winona Ryder | Fon: *49 621 72739834
Nordisch by Nature |  How to make an American Quilt | Fax: *49 621 72739835

Hi, Marc!

On Mon, May 02, 2005 at 07:20:29AM +0200, Marc Haber
wrote:
> On Mon, May 02, 2005 at 02:30:08AM +0300, Alexander Gattin wrote:
> > I just forgot about a "secret" -O option in
> > useradd. :)
> > You can use it to override whatever values from
> > login.defs
> > 
> > For example, regarding bug #298883, use of:
> > > `useradd -O PASS_MAX_DAYS=99999 usr298883`
> > will solve your problem (other solution, that you have
> > currently implemented, involves `chage`).
> > 
> > Thus closing the bug.
> 
> Reopening. Please document this option. I cannot use a secret,
> undocumented option.
Actually it is documented, although in TFS. ;) It''s not
a passage about TFS being self-obvious :), there''s
really a comment by Tomasz.

Anyway, I don''t object reopening the bug, especcially
because we have a problem with -O/-K (more on this
below). And also because I agree that it''s better to
have a documentation in "binary distribution" form,
i.e. in man/info/html/README/whatever pages (just pick
any you like ;)).
> > The problem with -O option is a Tomasz''s desire to
> > rename it to -K for sake of unfication with Solaris:
> 
> So we can/need to wait until it is renamed, then document it,
This depends on you.
Currently I see the next in adduser:
> Depends: ... passwd (>=961025)
Dependency on such an antique version of passwd looks
strange for me (do you really test adduser against
it?), but if you want to choose as old a passwd as
possible for the dependency, then probably you should
use chage instead of -O/-K, because chage is in passwd
since ages (a working version since 970616-1, AFAIU?).

I may check when the -O was introduced. I think it''s
too a very old option, and if you choose to start
relying on "-O" when we document it, you can still
declare dependency on an old passwd (I think about
2000xxxx). Of course, upcoming switch to "-K" can force
us to support _both_ options for some time, but it''s
not a big problem.

If you choose to wait for "-K" being introduced and
documented, then be prepared for long wait. First we
need to sync with upstream (4.0.8 currently), and
after that it''s still not clear when Tomasz will switch
to "-K". Moreover you will need to make dependency on
the very new pesswd, smth. like:
> Depends: ... passwd (>=1:4.0.10-1)
-- 
WBR,
xrgtn

Hi,

On Sat, May 07, 2005 at 09:58:19PM +0300, Alexander Gattin
wrote:
> On Mon, May 02, 2005 at 07:20:29AM +0200, Marc Haber wrote:
> > Reopening. Please document this option. I cannot use a secret,
> > undocumented option.
> 
> Actually it is documented, although in TFS.
TFS is not in the binary package, hence the binary package is missing
the docs. That''s a policy violation *ducks*
> > > The problem with -O option is a Tomasz''s desire to
> > > rename it to -K for sake of unfication with Solaris:
> > 
> > So we can/need to wait until it is renamed, then document it,
> 
> This depends on you.
Well, I decided on using chage in adduser, so the connection between
this bug and the adduser issue isn''t there any more. However, I find
it quite ugly to leave useful options undocumented. But that''s of
couse your decision.
> Currently I see the next in adduser:
> > Depends: ... passwd (>=961025)
> 
> Dependency on such an antique version of passwd looks
> strange for me
There surely was a reason for doing so, but since this dependency is
satisfied since bo (which had exactly this version of passwd), I think
that it can probably be removed.
> (do you really test adduser against it?),
No.
> but if you want to choose as old a passwd as
> possible for the dependency, then probably you should
> use chage instead of -O/-K, because chage is in passwd
> since ages (a working version since 970616-1, AFAIU?).
chage is now being used.

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don''t trust Computers. They | Mailadresse
im Header
Mannheim, Germany  |  lose things."    Winona Ryder | Fon: *49 621 72739834
Nordisch by Nature |  How to make an American Quilt | Fax: *49 621 72739835