On Wed, Aug 1, 2012 at 1:17 PM, Philip Martin <philip.mar...@wandisco.com>wrote:
> Stefan Fuhrmann <stefan.fuhrm...@wandisco.com> writes: > > > On Tue, Jul 31, 2012 at 11:26 AM, Philip Martin > > > >> I suppose we might want to > >> construct the full-text corresponding to the representation to verify > >> the checksum. > > > > > > Yes. Worst thing that might happen is a crash because > > we didn't check some buffer size. combined_window_cache > > will reduce the I/O overhead for frequently checked reps > > such as file properties. > > If this could crash then we want it to crash during the commit rather > than on subsequent access after an apparently successful commit. > +1. But there is a problem: it will (most likely) crash *every time* the user tries to commit the triggering content. So, we should put something like this into our docs: * if you see repeated crashes in commit, run svnadmin verify * if that crashes too, delete the rep-cache.db -- 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
