[Dnsmasq-discuss] ubus and metrics

[Simon Kelley]

On 23/04/18 23:52, Julian Kornberger wrote:
> Thanks for your comments.
> 
> I have added a `--enable-dbus` flag. It compiles but is not tested yet.

Thanks.

> 
> Where does the DBUS raise a fatal error if support is missing? I
> couldn't find anything in the code.

src/dnsmasq.c:

  if (option_bool(OPT_DBUS))
#ifdef HAVE_DBUS
    {
      char *err;
      daemon->dbus = NULL;
      daemon->watches = NULL;
      if ((err = dbus_init()))
        die(_("DBus error: %s"), err, EC_MISC);
    }
#else
  die(_("DBus not available: set HAVE_DBUS in src/config.h"), NULL,
EC_BADCONF);
#endif


> 
> General questions:
> - Is there any reason to not use the `enum` for definition lists like
> `LOPT_*`?
No strong ones, just habit I guess.

> - Is it planed to make the indentation consistent? Like only tabs or
> only spaces.
> 
I edit using emacs, and never see a problem. A massive edit would
generate a huge number of spurious changes in the git repository. I use
"git blame" quite often and don't want to find that it tells me half the
lines were last changed in the great re-tab.

Is there any advantage to doing it?


cheers,

Simon.


> Regards,
> Julian
> 
> On 24.04.2018 00:07, Simon Kelley wrote:
>> Can I make another request? It would be good (and consistent) to have a
>> dnsmasq option for UBus analogous to --enable-dbus, called --enable-ubus
>> so that dnsmasq doesn't try and connect to the Ubus unless that's
>> explicitly requested, even if the code is compiled in. Like the DBUS
>> version, it should raise a fatal error if the options is set but the
>> UBus code is not compiled in.