RFC: ABI break

[Robert Millan]

On Fri, Dec 30, 2005 at 10:18:09AM +0100, Aurelien Jarno wrote:
> On Fri, Dec 30, 2005 at 09:41:06AM +0100, Robert Millan wrote:
> > On Thu, Dec 29, 2005 at 09:09:59PM +0100, Aurelien Jarno wrote:
> > > > Expected etch release is after 12 months,
> > > > these days might be kernel of FreeBSD 6.x the preferred one.
> > > 
> > > Maybe yes, but currently it is not yet usable, so doing changes that may
> > > break support for FreeBSD 5.4 give me some fear. If we could have a
> > > FreeBSD 6.x that proves to work well, I am not against making such
> > > changes. That's why I prefer such changes to be done later.
> > 
> > Why don't we just take the same approach as upstream?  If upstream considers
> > 6.x stable enough to deprecate 5.4 completely, then so do we?
> > 
> Well, I would prefer to do that only if also 6.x is stable enough for
> us. Currently that doesn't seems to be the case as openpty() does not
> work, the userland tools (kldutils for example) have not been ported or
> well tested in version 5.4 on a 6.x kernel. Moreover the 6.x kernel has
> been tested by very few persons, maybe it brokes some things.

kldutils from 5.4 work fine, I tested them.  When the openpty problem is
addressed, we can provide 6.0 packages as an alternative, but not default.

My point is that when upstream stops maintaining 5.x (even for security updates)
it'll be unviable for us to keep with it.  This will probably happen when 7.0
is out, and I think we should move at that time (nor sooner neither later).

-- 
Robert Millan