Re: Encrypted Repositories. . .?

Karl Fogel wrote:
> Greg Thomas <thomasgd@omc.bt.co.uk> writes:
>> Well, I can see a way it could be. svn is enhanced such that the code
>> is encrypted at the client side, and then passed over the network and
>> stored on disk, still encrypted. When you check out the code, it's
>> decrypted client side.
>>
>> The drawback is that all files would essentially become binary and
>> hence non-mergable.
>
> No, merging happens on the client-side anyway, so the files would
> still be mergeable.

I thought the diffs were computed on the server side and applied to the
client files in a merge.

> In fact, outside the authentication step for the
> encryption, this scheme should result in no user-visible behavioral
> changes on the client side.

What would you expect 'svn diff ...' to show? And the deltas stored for
each version would be as large as the file itself.

> On the server side, though, repository hooks would no longer be able
> to examine the contents of files. That could be a problem.

A version control system that can't do efficient diffs will just be a
clunky backup system that makes a full snapshot copy for every change
(which could be done into an encrypted file if that's what you want).

-- 
   Les Mikesell
    lesmikesel@gmail.com
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org