Re: [patch rfa:doco rfc:NEWS] mi1 -> mi2; rm mi0

[Daniel Jacobowitz <drow@mvista.com>]

On Mon, Sep 30, 2002 at 11:03:55AM -0400, Andrew Cagney wrote:
> 
> >>>
> >>>Are you planning to revert mi1 then?
> >
> >>
> >>Que?
> >
> >
> >"mi2" changes have been sneaking in.  Are you planning to revert them -
> >create an "mi1" which matches what mi1 actually was.
> 
> It's a bit late for that.  Someone should audit the changes made so far 
> and identify which caused syntax changes and update accordingly.  Fixes 
> could, perhaphs be pushed into 5.3 (but I don't have the time).
> 
> >Otherwise, where is the line drawn to mark the interface version as
> >final?  It seems to me that the default shouldn't be evolving, that
> >-i=mi should default to a fixed point until the next version is
> >running.
> 
> I think a line is drawn when each release is made.  I'd expect an MI 
> client to explicitly specify -i=miN (where N was formally released) 
> rather than trust -i=mi.
> 
> However, should the HEAD hold off on recognizing -i=mi2 until the next 
> branch is cut?  On the HEAD, -i=mi evolves by definition.  However, 
> -i=mi2 is evolving as well :-(

That'd be best I think.  I think that -i=mi2 specifies a fixed standard
and we don't have one yet; so how about -i=mi being different from
-i=mi1, but not adding -i=mi2 until we're ready to fix the interface?

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer