At 4:38 PM -0500 8/14/03, Wadsworth, Eric (Contractor) wrote:
>So, yes, everyone can check out a new working copy, but then they have to
>manually merge their changed files into it... or the admin has to gather all
>the best bits from the working copies, and manually merge them in... you see
>the mire here.
I do. And I think your trick of shoving post-backup changes up from
the working copies has some definite charm. But it really is, as
has been suggested, a long road on which you embark.
And it's not at all guaranteed to completely work! Certainly the
history will likely not all be recoverable (backup at revision 1000,
various wc's available at revisions 2000, 2500, 3000; lots of
revisions missing, there! We'll never again know who committed which
at any of those). Very probably, not all the HEAD versions will be
recoverable (oh, sure, doubtless someone has a full trunk/* wc of
something darned close to HEAD, but what about tags/* and
branches/*?). You will say "better than nothing," and I won't argue.
But what's the return, what's the investment, what's the ROI?
I've done this kind of repository recovery for really large sites
(terabytes). In some cases, I've done it using ClearCase MultiSite
(have a crash, three days later you've only lost ten minutes' work
;-(). I believe in the value. But I'd like us to spend some serious
time thinking about maximizing the return before we jump into a
decision, let alone a design. This is *much* bigger than, say,
cvs2svn (and, oh, look, there's Karl grovelling on the floor at the
mere mention!;-).
--
-==-
Jack Repenning
CollabNet, Inc.
8000 Marina Boulevard, Suite 600
Brisbane, California 94005
o: 650.228.2562
c: 408.835-8090
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Fri Aug 15 00:04:08 2003