Re: [DIFFS] Re: Connection to emacs CVS broken ?

[Miles Bader]

Nick Roberts <address@hidden> writes:
>  > > Maybe the best thing is to restore from NIIMI Satoshi's rsync of the
>  > > repository?  [He said it's from 2007-03-13 02:48 GMT]
>  > 
>  > Indeed!  This would restore the CVS files, not just the changes.
>
> I don't know NIIMI Satoshi, but I'm sure he's a fine fellow.  However, it's 
> one
> thing to rely on an outside source for the last few changes, and another to
> rely on such a source for the *entire* revision history of Emacs.
>
> If we do follow this route then I would urge caution and suggest that someone,
> perhaps from Savannah, check that the only differences are, indeed, only in 
> the
> last few changes.

Right.  That's why I said:  "I think the changes from the restored
repository to the rsync would be small enough that they could probably
be verified by eye."

What I had in mind was simply doing a diff of the raw restored
repository tree to the rsync repository tree.  CVS files are fairly
understandable text format, and the number of changes on the trunk is
small enough that I think simply looking at that diff should be good
enough for a small number of trunk changes.

Unfortunately, now that I've thought about it more, a potential problem
comes to mind:  I think my last trunk->unicode sync is missing from the
restored repository, and is probably in the rsync'd repository, and
_those_ changes are not small.  Since CVS mixes all branches in a single
CVS file, interpreting those parts of the diff might not be so nice (and
it affects a lot of files)...

-Miles

-- 
Yo mama's so fat when she gets on an elevator it HAS to go down.