MD4 checksum fix

[Donovan Baarda]

On Tue, 2003-04-01 at 13:34, jw schultz wrote:
> On Tue, Apr 01, 2003 at 12:41:39PM +1000, Donovan Baarda wrote:
> > On Tue, 2003-04-01 at 09:35, jw schultz wrote:
[...]
> Comparative complexity is a matter of perspective.  I think
> it is a bit more deterministic.  It has been a few years

I was talking in simple LOC what-the-hell-does-this-do terms. As this is
not heavily exercised (once per file), it doesn't really matter.

> > I'm not particularly interested in supporting the old rsync/librsync
> > mdfour API.
> 
> I can understand that.  I do think the API change and the md4
> bug fix should be done separately.  The API change is really
> internal, it is the bug fix that affects the protocol.

I'm not 100% sure how close the current mdfour librsync API is to the
rsync API. If they are the same, you should just be able to drop it in
as a replacement, and grab the broken implementation of rs_mdfour_tail
to make a new rs_mdfour_broken_result

Hmmm... if someone else was going to do all the rest of the integration
work, I _could_ do this bit. I'd rather do it for the RSA API, but it
wouldn't be too much work to change this bit over later on.
 
> All i'm offering here is to coordinate with someone fixing
> the bug or to take the fix (if it can be agreed upon) from
> someone who can encapsulate it into a patch.  And that just

I don't really have time to do this myself, and don't have the
familiarity with the wider rsync code to do it. I've been focusing on
librsync (and will continue to do so).

-- 
----------------------------------------------------------------
Donovan Baarda                http://minkirri.apana.org.au/~abo/
----------------------------------------------------------------