On Thu, Oct 30, 2003 at 02:34:14AM -0500, Chris Ricker wrote:
On Wed, 29 Oct 2003, Michael K. Johnson wrote:
> I think that the fact that SMIT had to do this was an indication that
> the whole configuration process was wrongheaded from the start in AIX.
> I have always felt that keeping a database and then overwriting a set
> of functionally write-only config files was the worst combination of
> possibilities available. (Yes, I was an AIX admin at one time, I'm
> not just spouting off.)
I'm curious about your take on redhat-config-printer, then ;-)
As I said, I haven't been responsible for config tools at Red Hat for
years. Yes, Crutcher and I had arguments about redhat-config-printer.
However, there's one issue we had there that really overrides my
arguments: until FC1, we were trying to support two very different
print systems from one tool, including migrating configurations.
So it's not necessarily the wrong decision in this particular case.
At least it doesn't regen the config files every time you sneeze,
and you *can* choose to configure by hand if you like and not have
your files overridden at all; that's a huge difference from SMIT.
I'll agree AIX configuration sucks. One thing SMIT does which is
though. You can have it show command-lines which it is running behind the
scenes. Some aspects of SMC (Sun's java admin crapplet) are the same way.
That's a nice feature in GUI admin tools for newbies b/c it lets them slowly
learn the right way to do something if they want to migrate to the command
But so much is NOT "run this command" but instead "edit this configuration
"He that composes himself is wiser than he that composes a book."
Linux Application Development -- Ben Franklin