Stefan Sperling wrote: > On Thu, Nov 01, 2012 at 05:22:14PM +0200, Daniel Shahaf wrote: >> +1 to having a non-zero exit code if there was any error throughout. > > I think we should try to keep this discussion within the scope > of the current patch, which is about adding a --keep-going option > and providing sensible output when it is used, in order to > keep things simple for Prabhu. We can build upon that later > and defer things like munging the exit code to a separate patch.
I would be very reluctant to commit any patch that allows "svnadmin verify" to return a zero exit code after it found an error. Keeping the scope of the change constrained is a worthy goal but the possibility that the required follow-up might get forgotten is too much. I would want the second patch to be ready to commit before we commit the first. At the very least there must be agreement on how it's going to exit and commitment to make it happen. I already asked the question several messages back. Prabhu, please share your thoughts. - Julian
