Re: [Monotone-devel] Re: Fwd: Re: FreeBSD's requirements for its future VCS

[Brian May]

>>>>> "Nathaniel" == Nathaniel Smith <address@hidden> writes:

    Nathaniel> The two big questions about obliteration are: 1) Do you
    Nathaniel> want the data to simply be unavailable, or do you want
    Nathaniel> to remove every trace of its existence?  Approaches
    Nathaniel> like yours that leave the file hash in the manifest
    Nathaniel> mean that even if the file itself is gone, it is always
    Nathaniel> cryptographically provable that it _once_ was
    Nathaniel> available.  Conceivably, you might be obliterating a
    Nathaniel> file to preserve trade secret protection or to
    Nathaniel> otherwise protect yourself legally, and leaving this
    Nathaniel> kind of trail behind would be bad.  2) How smooth does
    Nathaniel> the obliteration process have to be?  Is it acceptable
    Nathaniel> if mirrors require manual intervention?  Is it
    Nathaniel> acceptable if workspaces have to manually recreated?

I think to answer these you would need to know why obliteration is
required in the first place.

I can think of two evil reasons:

* somebody committed data without permission and copyright owner is
jumping up and down threatening legal action if it isn't removed
immediately.

* somebody accidently committed passwords or other secret information.

Yes, neither of these should happen in the first place, but what if it
did happen?

So I suspect removing the trail my be required, but this is going to
be evil, and potentially break past revisions - it might just about be
better to start a new archive from scratch.

Also, what happens if only one line of one file is under dispute? Or
if the dispute is resolved and the copyright owner agrees the code can
stay after it has already been obliterated?
-- 
Brian May <address@hidden>