Re: PAM release status
On Sun, 4 Apr 2004 23:16:39 -0500
Steve Langasek <vorlon@debian.org> wrote:
> > > The behavior without the solution is difficult for users to
> > > understand.
>
> > You're saying that *conffile* prompts are difficult for an admin to
> > understand? Are you sure these folk are using Debian?
>
> It is difficult for a user to understand why they're being asked about
> whether they want to keep their locally modified copy of a config file
> they never modified.
>
> For some users, it's not difficult, because they know enough to
> recognize this as the result of a serious-severity bug; but woody
> shipped with that bug, so the damage is done.
Well, I certainly won't argue that having conffile prompts is worse than
having everything working perfectly.
However, if the admin somehow manages to get through an entire
stable->stable upgrade without enough conffile prompts to at least get
them *comfortable* with them (I remember my first stable->Sid, I got at
least a halfdozen), then I'd be shocked.
So all I'm saying is that "conffile prompts confuse the admin" shouldn't
be a particularily heavily-weighted bit of information unless a) it
proves to be true, and b) the admin can function fine without an
understanding of said prompts. Far more compelling would be "there will
be so many needless conffile prompts as to make the upgrade downright
painful." Looking at my own Woody machines, this doesn't appear to be
the case, and as I said I'm perfectly happy with getting conffile
prompts - Just In Case(tm).
That being said, I will admit that I might not care as much about other
conffiles. But we're talking about /etc/pam.d/, which is a pretty bad
place to special-case (didn't we learn our lesson the first time?) and
an even worse place to try and do something tricky.
--
Arguing with an engineer is like wrestling with a pig in mud.
After a while, you realise the pig is enjoying it.
OpenPGP v4 key ID: 4096R/59DDCB9F
Fingerprint: CC53 F124 35C0 7BC2 58FE 7A3C 157D DFD9 59DD CB9F
Retreive from subkeys.pgp.net or risk key corruption
Reply to: