introduce nscd --foreground, for systemd use

[Thomas Schwinge]

Hi!

I agree that this functionality is a useful addition.


On Mon, 30 Jan 2012 23:12:29 -0200, Alexandre Oliva <aoliva@redhat.com> wrote:
> On Jan 23, 2012, Roland McGrath <roland@hack.frob.com> wrote:
> 
> > This seems generally sound to me.  But I'd like to know if there is
> > any existing precedent for an option letter and/or long option name
> > for doing this in another GNU utility.
> 
> I'm afraid I couldn't find examples of programs that offered this level
> of flexibility, so I saught guidance in the GNU coding standards.
> http://www.gnu.org/prep/standards/standards.html#Option-Table
> 
> ‘foreground’
>     For server programs, run in the foreground; in other words, don’t do
>     anything special to run the server in the background.
> 
> ‘debug’
>     ‘-d’ in make and m4; ‘-t’ in Bison.
> 
> There doesn't seem to be a standard short option for “--foreground”;
> indeed, some non-GNU programs use -d to daemonize, whereas others use it
> to enter debugging mode.  I couldn't tell (by just looking at remote
> docs) whether they changed behavior WRT syslog when they offered both
> options.  It's a bit of a mess already :-(

Why do we need a short option at all?  Aren't these mainly for saving a
few keyboard hits in interactive shell use?  Unless really established, I
tend to prefer using long options in anything that's scripted; it's more
clear to the reader, as you have just demonstrated (-d vs. -t, for
example).

> > Perhaps cleaner to have two bools.
> 
> I've considered adding --[no-]syslog, but we don't really have 4
> different possibilities in the choice space.  It doesn't make sense to
> send output to std* if you daemonize: it's really 3-state.

What about using an enum?

> >> +	fprintf (stderr, _("further output sent to syslog\n"));
> 
> > What's the point of this message?
> 
> It's just so that someone who runs the program in --foreground isn't
> surprised by the absence of output when the program runs in foreground,
> behavior that is new and unlike that of -d.

I agree this message should go away.  ``No message == all OK'' is the
standard UNIX paradigm.  If I were systemd, I would potentially consider
seeing a daemon write to stderr as an indication for an error.

> > It's not normal for daemon startup to print anything to stderr in my
> > experience.
> 
> This won't be printed in daemon mode, only in --foreground mode.


Grüße,
 Thomas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 489 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/libc-alpha/attachments/20120301/a1861f9c/attachment.sig>