conf/104884: Add support EtherChannel configuration to rc.conf

[Brooks Davis]

On Thu, Feb 08, 2007 at 11:48:04PM +0000, Florent Thoumie wrote:
> Brooks Davis wrote:
> > On Thu, Feb 08, 2007 at 11:30:41PM +0000, Florent Thoumie wrote:
> >> Brooks Davis wrote:
> >>> On Thu, Feb 08, 2007 at 01:55:16PM -0800, Doug Barton wrote:
> >>>> Brooks Davis wrote:
> >>>>
> >>>>> The default should be an empty list which results in nothing happening.
> >>>>> I'd suggest making empty list the value for the default gif_interfaces
> >>>>> in /etc/defaults/rc.conf in both branches, removing support for NO in
> >>>>> CURRENT and emitting a warning in stable.
> >>>> How about issuing a warning for NO in both branches? Whether I agree
> >>>> with you or not on the importance of keeping things clean and
> >>>> consistent, I definitely do not want to err on the side of pedantry
> >>>> over usability.
> >>> That would be fine.  I don't really care as long as it's deprecated.
> >>>
> >>> FWIW, only users who don't update /etc/defaults/rc.conf or who manually
> >>> set gif_interfaces="NO" would be effected so the size of the set of
> >>> effected users is probalby close to epilon and even all that will happen
> >>> is cloning an extra interface and then not configuring it so it should
> >>> be basicly harmless to just remove direct support for it.
> >> Fine with me as well. Should we make it a warning on RELENG_6 and an
> >> error on HEAD, or a warning on both. The former being be what I was
> >> planning to do, ie. remove support for "NO" in HEAD but issue a message
> >> saying semantics have changed. The latter would mean identical code in
> >> both HEAD and RELENG_6 (so "NO"-compatibility in both branches), but
> >> we'd need a reminder to remove this "NO"-support in HEAD once RELENG_7
> >> is branched.
> > 
> > I'd say a warning in both.
> 
> Re-reading Doug's message, he's probably thinking the same thing, but
> this is for gif_interfaces only, right?

That's what I'd do.  There's no reason to introduce support for an
instantly deprecated feature in a new variable, particularly since
gif_interfaces is the odd one out.

-- Brooks