On Tue, Aug 03, 2004 at 10:26:02PM -0400, Greg Hudson wrote:
> On Mon, 2004-08-02 at 09:04, Chia-liang Kao wrote:
> > If you load the attached dump into bdb, you will see something like
> >
> > A /foo/bar.cp2 (from /foo/bar:2)
> >
> > in the log. But with fsfs, you will see
> >
> > A /foo/bar.cp2 (from foo/bar:2)
>
> Fixed in r10480 and nominated for backport to 1.1. I only fixed the
> problem on the way into the changed-path data, not on the way out;
> existing repositories where non-canonical copyfrom paths have crept into
> the changed-path entries will have to be dumped and loaded in order to
> be fixed.
>
> (The problem seems fairly harmless to me; if it's a big deal, we could
> add code to canonicalize copyfrom data as it is read from the
> changed-path entries. Certainly, I would have done that if we had
> already hit 1.1 with this bug, or if the problem had severe
> repercussions.)
>
This may be a non-issue, but the blame history-traversal code used to
pay attention to the paths. I haven't looked at it recently,
though. Either way, a dump/load would fix the problem, if any.
--ben
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Aug 4 16:56:32 2004