On Tue, Jul 31, 2012 at 10:45 AM, Daniel Shahaf <danie...@elego.de> wrote:
> Stefan Fuhrmann wrote on Tue, Jul 31, 2012 at 10:35:24 +0200: > > On Thu, Jul 26, 2012 at 11:41 AM, Philip Martin > > <philip.mar...@wandisco.com>wrote: > > > > > Stefan Fuhrmann <stefan.fuhrm...@wandisco.com> writes: > > > > > > > Yesterday, I debugged the code and found out why r(N-2) > > > > would be reported. This was due to is-fresh-txn-root > > > > being set on some of the root noderevs. Some of the > > > > affected repositories don't use directory deltification. > > > > Maybe, I'm able to look deeper into how that might > > > > have happened. > > > > > > > > I think we found another form of corruption. Since the > > > > fix is simply to ignore the flag, the question is whether > > > > we may ignore it during (de-)serializing noderevs. > > > > > > One way to set is-fresh-txn-root is to commit an empty rev: > > > > > > http://subversion.tigris.org/issues/show_bug.cgi?id=4031 > > > > > > > That fully explains why I see those flags in my repos. > > Now, svnadmin verify will report them as corrupted. > > Could svnamdin report them via warnings rather than via fatal errors? > They are harmless after all so I think they shouldn't mask "real" > corruptions. > Feel free to relax your check to support existing repositories ;) But I think we should also fix the root cause in trunk for 1.8. -- Stefan^2. -- *Join us this October for Subversion Live 2012<http://www.wandisco.com/svn-live-2012>– 2 full days of training, networking, live demos and more! 25% off before Aug. 10th with discount code “earlybird.” *Certified & Supported Apache Subversion Downloads: http://www.wandisco.com/subversion/download
