cvs commit: src/etc network.subr

[John Baldwin]

On Thursday 29 March 2007 03:08:12 pm Andrey Chernov wrote:
> On Thu, Mar 29, 2007 at 08:24:09PM +0300, Mike Makonnen wrote:
> > On Thu, Mar 29, 2007 at 10:55:34AM +0000, Andrey A. Chernov wrote:
> > > ache        2007-03-29 10:55:34 UTC
> > > 
> > >   FreeBSD src repository
> > > 
> > >   Modified files:
> > >     etc                  network.subr 
> > >   Log:
> > >   Fix get_if_var() with 3 args (i.e. with default)
> > >   
> > >   All xxx_<ifname> flags are set to empty strings automatically earlier 
so
> > >   eval echo \${${prefix}${_if}${suffix}-${_default}}
> > >   not substitute the default but return just the empty string.
> > >   Fix it using
> > >   eval echo \${${prefix}${_if}${suffix}:-${_default}}
> > >   (i.e. treat empty strings as unset)
> > >   
> > >   The bug manifistates itself with the following warning from 
checkyesno():
> > >   /etc/rc.d/dhclient: WARNING: $background_dhclient is not set 
properly -
> > >   see rc.conf(5)
> > >   
> > >   Revision  Changes    Path
> > >   1.178     +1 -1      src/etc/network.subr
> > 
> > Please back this out.
> > 
> > Your change doesn't fix the bug. It only hides it. The bug was fixed in
> > rc.d/dhclient v1.28. 
> 
> No, the bug was not fixed. It IS in v1.28

The change in v1.28 fixed a similar warning where you'd get something like

/etc/rc.d/dhclient: WARNING: $background_dhclient_bge0 is not set properly

(where bge0 is an interface name).  1.28 fixed the warning for me, but I don't 
use background_dhclient and have none of the variables set.  Maybe Andrey is 
seeing a warning in the non-default case when a variable is set?

-- 
John Baldwin