Re: Debdiff project announcement.
Je 2000/02/04(5)/ 0:02, Tom Rothamel montris sian geniecon skribante:
> This is just a message to announce that I am undertaking a project I
> call debdiff.
I've long thought that the diffs created by xdelta are way too big.
I have actually started writing a binary diff utility that created
diffs that were about 3 to 10 times smaller than xdelta (all compressed
sizes), but abandoned when I had less time, and the stuff became
But, I do have more time now (well, should have), and I did
actually plan to restart working on that diff tool.
Would my diff tool be of any help to you? I should be
able to send you sources in a couple of days, if you so desire.
Note: The `3 to 10 times smaller (compressed sizes)' means this:
I diffed the uncompressed .ar file inside a .deb.
the diffse created by xdelta are compressed by default, my
utility didn't compress them by default. But I noticed that
bzip2 was *much* more efficient compressing the delta's I
created than gzip, so I made xdelta create uncompressed
diffs, and then bzip2-ed both of our diff's. And with that
I remember my diffs were at least 3 times smaller (maybe 10,
don't remember any more).