[U-Boot-Users] Please pull u-boot-mpc83xx.git mpc83xx branch into testing

[Kim Phillips <kim.phillips@freescale.com>]

On Mon, 24 Sep 2007 16:47:58 -0600
"Grant Likely" <grant.likely@secretlab.ca> wrote:

> On 9/24/07, Kim Phillips <kim.phillips@freescale.com> wrote:
> > On Mon, 24 Sep 2007 23:52:29 +0200
> > Wolfgang Denk <wd@denx.de> wrote:
> >
> > > In message <20070924161309.9f14b16e.kim.phillips@freescale.com> you wrote:
> > > >
> > > > > There is no merge window open at the moment.
> > > > >
> > > > ok, let me understand how you want this to work.
> > > >
> > > > my original intent was for you to pull 837x support into your /testing/
> > > > tree so that gcl's mass-rename scripts could do their magic on the 837x
> > > > board port too.
> > >
> > > No, that will not work. You would move Grant's base under his feet.
> >
> > this goes both ways.  True, Grant's work is more intrusive, but that
> > just means that all work that is or can be affected by it be staged in
> > a single common tree.  My understanding is you have designated the
> > -testing tree for this particular purpose.
> 
> I don't think that's quite true.  In the last merge window, -testing
> got used to stage changes for the merge window with the least possible
> impact.  In that case, it was Jon's CMD_* changes which needed to go
> in first.

Jon's CMD_* changes were performed in stages, with boards being added
in between.  Sure, after each phase, things inevitably broke, and they
were promptly fixed up again, by him and others.  Isn't that how your
Kconfig changes are going to work (in phases)?

> It makes total sense to stage changes in the custodian trees, but
> Wolfgang retains the right to decide what order those changes get
> staged into -testing, and also to request custodians to
> merge/rebase/retest before pulling their tree.

sure, and I can do that, I just wasn't aware of existing blockage.  I
would have expected blockage to occur between the time you submit
patches to the mailing list and when WD applies them to his tree.

I just want your (and others') /subsequent/ patches to include 837x
code, so I, and other custodians btw, don't have to duplicate your work
unnecessarily and keep on having to play catch up.  But it sounds like
you're not going to do it in phases, or, at least the phases are going
to be clearly marked with releases 1.3.[123]?

> > > > but it sounds like you want me to hold on to new 83xx boards/features
> > > > until you make a final (non-rc) release?  Sounds like I should be
> > > > asking gcl to share his scripts then.
> > >
> > > No again. Please just wait until we have a somehwat  stable  base  in
> > > testing,  and Grant has applied his patches there. Then use *this* as
> > > base for any new submissions, which you then should post in the merge
> > > window.
> >
> > It sounds like you're suggesting everyone halt all new development
> > indefinitely.  Please clarify.
> 
> Not at all!  It is not a suggestion to halt development; but it is a
> warning that custodians need to do either a rebase or a merge before
> it is pulled when the window opens (or before Wolfgang chooses to pull
> them into -testing).

What most concerns me here is the indefiniteness of the situation.  I
think the process can be made more granular (and therefore efficient,
without WD having to deny pull requests), but I'll wait and see how
things go.

> Besides; if I come up with a set of changes that will make a merge or
> rebase absolutely impossible, I'm sure Wolfgang will tell me what I
> can do with my patchset.  :-)

yeah - we'll see :)

Kim