Re: state of the 2.6.13 transition

On Thursday 22 September 2005 02:47 pm, Erik Hovland wrote:
> On Thu, Sep 22, 2005 at 11:03:32PM +0200, Patrick Ohly wrote:
> > On Wed, 2005-09-21 at 00:30 +0100, Phil Blundell wrote:
> > > On Tue, 2005-09-20 at 16:14 -0700, Matthew Reimer wrote:
> > > > One more detail: I think the breakage relates to files that exist
> > > > only on the vendor branch; i.e., files that we haven't modified from
> > > > upstream.
> > >
> > > Yeah, exactly.  This all stems from the fact that CVS only tracks
> > > changes on a file-by-file basis: it doesn't have any understanding of
> > > the relationships between different files.
> > >
> > > CVS considers the HEAD to be a derivate of the vendor branch, which is
> > > reasonable enough in itself. Unfortunately, the way it implements this
> > > is that, for files that have never been modified in our tree, the tip
> > > of the vendor branch "shows through" into the HEAD.
> >
> > Would it be possible to fix this by creating a new branch
> > starting with the last known good version, then
> > making that branch the default? Something like:
> >   cvs update -d -r K2-6-12-hh3
> >   cvs tag -b NEW-TRUNC
> >   cvs admin -bNEW-TRUNC
> >
> > Just a thought...
>
> Or we could check out two trees, HEAD and K2-6-12-hh3. Diff between the
> two. Apply the diff to the HEAD tree and commit. That should clear it
> up.

FYI, Ian's almost done merging. We figured out what happened to the missing 
files (cvs refused to import files or directories named 'core', including 
net/core, usb/core and sound/core), so hopefully he can commit really soon 
now.

Matt