This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Ghost testcase eb42.C
- From: Ian Lance Taylor <ian at wasabisystems dot com>
- To: Daniel Jacobowitz <drow at mvista dot com>
- Cc: bkoz at redhat dot com, "Joseph S. Myers" <jsm at polyomino dot org dot uk>, Andrew Pinski <pinskia at physics dot uc dot edu>, gcc at gcc dot gnu dot org
- Date: 12 Aug 2004 20:05:36 -0400
- Subject: Re: Ghost testcase eb42.C
- References: <0201B623-BE44-11D7-9500-000393A6D2F2@physics.uc.edu><Pine.LNX.4.56.0307251112490.20499@kern.srcf.societies.cam.ac.uk><20040420223840.GA22662@nevyn.them.org>
Daniel Jacobowitz <drow@mvista.com> writes:
> On Fri, Jul 25, 2003 at 11:21:15AM +0100, Joseph S. Myers wrote:
> > On Thu, 24 Jul 2003, Andrew Pinski wrote:
> >
> > > This looks like it has been messed up for a long time.
> > > The last time it was touched was 2 years and 3 months ago but it was
> > > messed
> > > up before that from reading the log:
> > > <http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g%2b%2b.old-
> > > deja/g%2b%2b.robertl/Attic/eb42.C>.
> > > The reason I say it was messed up before that was from log about the
> > > merges into the branches.
> >
> > This sort of thing has been observed before - a file somehow getting
> > committed to after it has been removed, and the new revisions being in the
> > Attic but not marked as dead so getting resurrected on branches. See e.g.
> > c-gperf.h. I don't know *how* the files got committed to after being
> > removed, but if you can reremove them on mainline (i.e., get a new
> > revision committed to the head of the ,v file, with no changes from the
> > previous one but state "dead") then that should fix the problem.
>
> This just bit me again - the version of CVS on sources doesn't
> generally check out this file, but my local CVS working on an rsync'd
> repository does.
I just removed the file testsuite/g++.old-deja/g++.robertl/eb42.C for
real. It was deleted in 1998, and then I think it was accidentally
recreated because it was modified in a working directory and the
conflict was resolved by checking in the modified copy rather than
delete it from the working directory. Then the file got out of synch
in the CVS repository somehow, and was in the Attic not marked as
dead. It was not checked out ordinarily, but was tagged for at least
some branches. I made the file consistent by briefly adding it back,
and then deleting it for real.
I will readd it if somebody thinks that is what we need. I'm pretty
sure it should be deleted.
Ian