Date: Fri, 4 May 2001 21:42:37 +1000 (EST)
From: Bruce Evans <bde@zeta.org.au>
To: Dima Dorfman <dima@unixfreak.org>
Cc: audit@FreeBSD.ORG
Subject: Re: {get,set}progname functions
Message-ID: <Pine.BSF.4.21.0105042117490.60028-100000@besplex.bde.org>
In-Reply-To: <20010502060949.23EFC3E28@bazooka.unixfreak.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 1 May 2001, Dima Dorfman wrote:
> Bruce Evans <bde@zeta.org.au> writes:
> > I think setprogname() should not be a stub. setproctitle() can be
> > used to change the process's name, so why should setprogname() be
> > weaker? Perhaps setproctitle() should affect __progname (or at least
> > the result returned by getprogname()), and setprogname() shouldn't
> > exist.
>
> setproctitle() and setprogname() are supposed to change different
> things. setproctitle() sets the name of the program as it should show
> up in ps(1). Things like sendmail use it to show their status (e.g.,
> "sendmail: accepting connections"). Conversely, setprogname() sets
> the program name as should be reported by err(3) and friends. I don't
> think the authors of sendmail have output like
>
> sendmail: accepting connections: some random error
>
> in mind when they write
>
> warn("some random error");
OK. They are logically separate, so it is simpler for them to have no
interactions.
> That said, you're right that setprogname() shouldn't be a stub. In
> the corrected path (below), it simply sets __progname. The fact that
> the pointer passed to it is kept for the duration of the program's
> life cycle is documented, so I think it's okay.
OK.
> > > +const char *getprogname __P((void));
> > I think this prototype should be formatted as:
> >
> > const char *
> > getprogname __P((void));
> >
> > like all the other prototypes for all the other functions whose return
> > type (name) is longer than 7.
>
> All the rest of the prototypes aren't formatted quite like that. E.g.:
>
> unsigned short
> *seed48 __P((unsigned short[3]));
>
> The placement of the '*' differs. Personally, I think the way you
> suggest looks better, so that's the way I wrote it in the new patch.
> But which one's right?
The others also have a 9-char indent if they don't have a '*'. The
'*' looks better when it's in the same column (before the function
name) for all cases, although it logically doesn't really belong
there.
> As mentioned before, a corrected patch is attached.
OK...
> --- /dev/null Tue May 1 20:04:30 2001
> +++ lib/libc/gen/getprogname.c Tue May 1 22:58:10 2001
> @@ -0,0 +1,14 @@
> +#ifndef lint
> +static const char rcsid[] =
> + "$FreeBSD$";
> +#endif /* not lint */
Use LIBC_RCS as well as "lint" here (see libc/gen/err.c; other files
mostly use LIBC_SCCS but I think we plan to change them, at least if
the id is not an sccsid).
> +
> +extern const char *
> + __progname;
The line break for indenting the variable name is a bit much.
Bruce
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-audit" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0105042117490.60028-100000>