Re: [Gnu-arch-users] Re: patch: automatic remove the corrupt revlib revision

[Stephen J. Turnbull]

>>>>> "Stefan" == Stefan Monnier <address@hidden> writes:

    Stefan> But IIUC rather than removing the revlib, it would be
    Stefan> better to re-snap the inodes because the revlib hasn't
    Stefan> actually been corrupted.  Right?

It seems to me you'd have to be pretty tricky to save much over
rebuilding the revlib, at least if you wanted to be sure.  If all the
inodes are invalidated (for the purpose of this heuristic), then
effectively you are unsure whether any of the files are corrupt, and
you have to diff (or checksum) them all, right?  So you'd end up with
a substantially smaller constant on the O(n-bytes-in-revlib) term for
local operations (checksum v. read+write), but you'd still have
horrible*O(n) for dealing with remote archives.

OTOH, if you're not going to bother checking, you can save time on
everything that looks at the revlib by simply ignoring the inode _all_
the time.  :-)

Quite possibly I'm missing something, but the logic isn't
straightforward (to me, anyway).


-- 
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
               Ask not how you can "do" free software business;
              ask what your business can "do for" free software.