Ben Collins-Sussman wrote:
>
> We could speculate, I guess. But my speculation is that the chances
> that he's found a genuine bug in our dump feature is low; this thing
> has been tested and evolving for 5 years. Occam's razor says that
> there's likely some simpler explanation.
>
> If we were getting lots of similar reports like this, we'd be
> panicking. But one mysterious report in five years? We're simply
> curious at this point, awaiting more data.
>
Sorry to say, I cannot provide the information equivalent to "grep -E
(Revision-number:|Node-path:|Node-action:|Node-copyfrom-)' dumpfile".
I thought I could obfuscate the paths in some but since there is > 65000
node-paths that is too cumbersome.
The thing is that I'm working for a bank that has certain rules about
giving out information about their source tree.
Would it suffice to give out the information about the path causing the
trouble + another path that it was copied from?
I don't know how to proceed from this.
I've solved the problem in a really ugly way for now: I've filtered out
the bogus path while restoring and then committed the lost data to HEAD.
This is not good since tags has changed behind the scene, but I think we
can live with that...
Regards,
Magnus
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Aug 29 18:08:41 2006