Re: cvs co -r netbsd-5 brings back some Attic files

[Alan Barrett <apb%cequrux.com@localhost>]

On Fri, 31 Oct 2008, Tom Spindler wrote:
> FWIW, even if I cvs up -dP those files appear.
> 
> pts/8 dogcow@burai: /src/src/share/termcap 3015 % rm tck
> pts/8 dogcow@burai: /src/src/share/termcap 3016 % cvs up -dP -r netbsd-5 tck
> cvs update: warning: tck was lost
> U tck
> pts/8 dogcow@burai: /src/src/share/termcap 3017 % cvs log tck|head          
> 
> RCS file: /cvsroot/src/share/termcap/Attic/tck,v
> Working file: tck
> head: 1.3
  [...]

There's something strange about that file.  It's in the Attic, and
revision 1.2 has "state: dead", so presumably the revision 1.2 commit
moved it into the Attic.  However, there's a more recent revision 1.3
woth "state: Exp".  I would have expected the file to be moved out of
the Attic at the time revision 1.3 was committed.

I almost always cvs update using the -D option, as in "cvs update
-D"yyyy-mm-dd hh:mm UTC" or "cvs update -r${branch} -D"yyyy-mm-dd hh:mm
UTC".  I have share/termcap/tck file in my -current tree as well as in
my netbsd-5 tree, presumably because of the strange state of the file.

Similar issues apply to other files identified in this thread: They are
in the Attic, but not marked as "state: dead".  I don't know how this
can happen using normal cvs commands; the normal way for a file to be
moved to the Attic is (as far as I know) always accompanied by a commit
that sets "state: dead".

I suspect that, somewhere in the past, files were manually moved to the
Attic without using "cvs delete" followed by "cvs commit".  If that is
the case, I think that we should fix them now.

--apb (Alan Barrett)