Re: Proposed new /etc/rc.d/drvctl script

[Paul Goyette <paul%whooppee.com@localhost>]

On Wed, 9 Apr 2025, Robert Elz wrote:

>    Date:        Tue, 8 Apr 2025 08:19:38 -0700 (PDT)
>    From:        Paul Goyette <paul%whooppee.com@localhost>
>    Message-ID:  <Pine.NEB.4.64.2504080816030.2859%speedy.whooppee.com@localhost>
>
>  | Allll of your comments appearto address "structural" issue, and
>  | all of them exist in the original modules script.
>
> Yes, and I have no idea about the latter, no reason to look, none of
> my kernels have "options MODULAR" in them, so module related anything
> is of no interest.   What it might need in its config file is completely
> unknowable to me.

ok.

>  | Unless the current stuff is actually broken,
>
> It didn't look to be, and I'm guessing that you have tested (are using)
> it already.

It's in production use on my machine with no issues.

>  | I am inclined to commit this version, with an invitation to you and
>  | other to address and structural improvements.
>
> The only issue with that is the format of the config file, once
> committed in one form, so people (potentially) start using it, it
> gets very tricky to ever convert it to something different.

The same swapping of <name> and <args> happens in the modules script
and hasn't been touched since NetBSD-8.  No reason I know of to be
different, and I've seen no comments/complaints on modules.

> I wouldn't be touching it, I have no reason to use it, and hence
> no easy/rational way to test any changes.   It needs to be done by
> someone who is actually going to be using the thing and can confirm
> that any changes actually work.
>
> But I could send you a revised version to test (before you commit it)
> if you'd like - but I'd only be guessing at what the actual needs are.
>
> For example, is there ever going to be a reason to have whitespace
> (or one of the shell operator characters ; | & < > ( ) (etc, if I
> forgot any) or quoting chars (' " \)) in any of the args that need to
> be passed to drvctl ?

I'm willing to test anything that doesn't risk my one-and-only.  But
I'm not going to audit all the others to keep them in synch.  The
current state is good enough and perfection isn't necessary.

> I'm assuming from the way it is now, that there would never be a literal
> \n intended to be included in an arg.

not as far as I can see.

>  | Thank you for not objecting to the concept of doing this.
>
> The concept looks fine, and it isn't going to bother anyone who doesn't
> enable it.

Right.

+---------------------+--------------------------+----------------------+
| Paul Goyette (.sig) | PGP Key fingerprint:     | E-mail addresses:    |
| (Retired)           | 1B11 1849 721C 56C8 F63A | paul%whooppee.com@localhost    |
| Software Developer  | 6E2E 05FD 15CE 9F2D 5102 | pgoyette%netbsd.org@localhost  |
| & Network Engineer  |                          | pgoyette99%gmail.com@localhost |
+---------------------+--------------------------+----------------------+