removing -A

[Richard Sharpe]

On Sun, 22 Jun 2003, Andrew Bartlett wrote:

> > > > Hmmm, but do we really want to do that? I like having short forms like 
> > > > that, and surely we could test that information presented works before 
> > > > writing it to the secrets file?
> > > 
> > > Short-forms of things are good, when you do something often - but this
> > > is a once-off command.  Anyway, the information probably would work, so
> > > that test isn't that useful...
> > 
> > I am sorry, that is just not true. What about the case where the admin 
> > gets the authentication information incorrect? We should be checking the 
> > info they provide to prevent the admin from having to hunt around for the 
> > reason for subsequent failure.
> 
> Sorry, I wasn't clear - yes, we should test the info before we start
> it.  Be warned that it's not quite as trivial as it should be (unless
> you want to turn wbinfo into something the size of 'net' just to do it).
> 
> What I meant was that we can't just keep -A and -a on the basis that we
> we can tell if the user stuffed up - given that they both take the same
> input parameter format.

But the point is that a test in the -A case that the details provided 
actually work cathes both problems:

1. The case where someone did wbinfo -A blah blah when they meant wbinfo 
-a, and

2. The case where someone did wbinfo -A blah blah and supplied the wrong 
blah blah.

Of course, the test is going to have to ensure that the blah blah that was 
supplied allows winbindd to do what it needs to do.

The only remaining case, it seems to me, that is not handled is where 
someone does wbinfo -A when they meant to do wbinfo -a and they also 
supply account details that are powerful enough to do what is needed but 
are not the ones they intended to supply.

Regards
-----
Richard Sharpe, rsharpe[at]ns.aus.com, rsharpe[at]samba.org, 
sharpe[at]ethereal.com, http://www.richardsharpe.com