I'm not talking about branches, I'm talking about releases.

I don't personally have time right now to make any involved changes
to the TCP code without a significant chance of breaking something.
so I want to make a minimal change now (in TOT and the 2.0 branch)
so that TCP will work better in LP64 kernels in the 2.0 release.

when someone has time to make the more involved change of using
fixed-size types in all the appropriate places in the TCP code,
then that person can decide whether those changes are desirable
to have in whatever release branches are relevant at that point.
I'm assuming that will be after the 2.0 release.  if someone has
time to do this before the 2.0 release, that would be fine, but
I haven't heard anyone volunteer.

-Chuck


On Tue, May 04, 2004 at 09:35:45AM +1000, Daniel Carosone wrote:
> On Mon, May 03, 2004 at 07:34:34AM -0700, Chuck Silvers wrote:
> > [...], but I'd rather do that after 2.0.
> 
> I know you know this, but it is already "after 2.0" as far as branches
> are concerned.  
> 
> This kind of change will make later tcp pullups to the branch harder,
> but that's still going to be true long after the 2.0 release is
> formalised and finalised, and indeed after the next release as well
> (since we typically maintain two prior branches).  So there's really
> no better or worse time to make the change on the HEAD [*].
> 
> I think what you're trying to say is "make this this smaller change
> first, and pull it up only", which is fine - but there is, just
> perhaps, an argument to pull up the "proper" change instead.  Or
> perhaps you're saying "pull up this small change before 2.0, and the
> larger change after the beta cycle"?
> 
> Anyway, mostly a terminology nit, because I've seen a couple of
> comments recently using "loose language" that may confuse some people
> about how the branches work - but also something to think about wrt
> what gets pulled up.
> 
> --
> Dan.
> 
> [*] Unless you're anticipating another set of tcp changes needing
> pullup shortly after. I have a funny feeling, after the recent TCP
> vulnerability issues, that we may see more TCP tweaks soon - and
> having the HEAD and branch in sync would be an advantage for that
> later maintenance.
>