Re: am I misunderstanding what HEAD refers to?

[Spiro Trikaliotis]

Hello Ted,

* On Fri, Oct 05, 2007 at 11:54:07AM -0700 Ted Stern wrote:
> On 05 Oct 2007 11:04:38 -0700, Dennis Jones wrote:
> >
> > "Ted Stern" <address@hidden> wrote in message 
> > news:address@hidden
´[...]
> >> So my first naive thought is to do a diff.  Note that I'm currently
> >> sitting in my branchname sandbox:
> >>
> >>    cvs diff -r BRANCHNAME_MERGE_HEAD -r HEAD 1>stdout 2>stderr
> >
> > I'm not sure why nobody eslse has suggested this, but since 
> > BRANCHNAME_MERGE_HEAD always refers to a point on the trunk and you want to 
> > diff the trunk, why not go to the trunk and do the diff from there?
[...]
> To do what you suggest, the user would have to do a 'cvs update -dA'
> [or 'cvs update -d -r parentbranch' if not branching from the trunk].
[...]

Ok, another try. From your description, I have not seen any reason why a
*r*diff would not work:

    cvs rdiff -r BRANCHNAME_MERGE_HEAD -r HEAD MODULENAME 1>stdout 2>stderr

IMHO, it would not work only if there would have been local changes. If
this might be an issue, an additional

   cvs diff 1>stdout.local 2>stderr.local

would do the trick.

This way, no additional "cvs up" would be needed.

Ted, am I missing something now?

Regards,
   Spiro.

PS: I reverted to using "rdiff" almost always now when I want to check
    for changes in the repository. "cvs diff" has the disadvantage that
    it ignored directories which are not available locally, while
    "rdiff" handles these, too.

PPS: Yes, I know that "cvs diff" intentionally works this way, and there
     are also uses where I want exactly *that* behaviour.
     Unfortunately, I have been hit by this behaviour more than once,
     missing important diffs between branches.

-- 
Spiro R. Trikaliotis                              http://opencbm.sf.net/
http://www.trikaliotis.net/                     http://www.viceteam.org/