On Thu, 2002-01-24 at 09:53, Torgeir Veimo wrote:
> I have experimented with the xdelta algorithm, using the gdiff file
> format. I guess the format is not that different from your homegrown
> svndiff. (http://www.w3.org/TR/NOTE-gdiff-19970901.html)
Not so different, except that gdiff isn't windowed, and svndiff is.
(You could certainly use a sequence of gdiffs with a little bit of
header information on each one, I suppose.)
> Did you find vdelta better than xdelta at generating compact diffs?
We didn't do a comparison. At the time we were making the decision,
xdelta was (at least, to the best of our knowledge) not library-ized
such that it would be useful to us; you had to bring in a bunch of other
stuff as well. Today, we would probably have to take a closer look at
it. I'm certainly curious whether it does a better job than svndiff
with equivalent windowing... except that I get the impression that
xdelta isn't windowed, and therefore wouldn't produce the nice streamy
behavior we can get with svndiff.
(You can search for "xdelta" in the mailing list archives as well, to
see the dozen and a half messages which discussed it.)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:36:59 2006