dist-hg proof-of-concept ready for use

[Carl Worth]

On Tue, 14 Nov 2006 07:18:55 -0500, Alan Cox wrote:
> Having seen git trash a repository and the fsck of git say nothing was
> wrong I'm firmly against adopting git for anything long term and serious
> until it is a lot more robust.

Alan, you've mentioned this trashing twice in this thread. Do you have
anything more than anecdotal evidence about this? If we are to take
your recommendation seriously, how do you recommend we measure the
robustness of git?

Even if there are bugs in the current git-fsck-objects, (something I
haven't seen), the git design for full-history integrity is extremely
compelling. Barring bugs, any corruption to the history can be
detected with an fsck, and no corrupt object can ever enter the
repository to replace a good object, (nor will the protocols allow an
object to enter the repository under a false pretense of its contents
since content is transferred, not hashes, which are always computed
locally).

And seriously, we're talking about a replacement for cvs here. In
every major conversion effort away from cvs that I have seen, the
primary difficulty has been dealing with corruption in the cvs data,
(both from users doing "bad things" sue to cvs limitations, and due to
other corruption that cvs never noticed).. I strongly doubt there is
any large, long-term cvs repository without serious damage already.

-Carl
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/fedora-maintainers/attachments/20061114/0809553b/attachment.sig>