[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gnu-arch-users] GNU Arch 2.0 and binary files
From: |
Jeremy Shaw |
Subject: |
Re: [Gnu-arch-users] GNU Arch 2.0 and binary files |
Date: |
Fri, 21 Oct 2005 13:57:56 -0700 |
User-agent: |
Wanderlust/2.15.1 (Almost Unreal) SEMI/1.14.6 (Maruoka) FLIM/1.14.7 (Sanjō) APEL/10.6 Emacs/21.3 (i386-pc-linux-gnu) MULE/5.0 (SAKAKI) |
At Fri, 21 Oct 2005 16:21:54 -0400,
Adrian Irving-Beer wrote:
>
> [1 <multipart/signed (7bit)>]
> [1.1 <text/plain; us-ascii (7bit)>]
> On Thu, Oct 20, 2005 at 05:25:07PM -0700, Andy Tai wrote:
>
> > Previously there were discussions of how to efficiently store binary
> > files, like using xdelta or such.
>
> I found "rzip" the other day:
>
> The principal advantage of rzip is that it has an effective
> history buffer of 900 Mbyte. This means it can find matching
> pieces of the input file over huge distances compared to other
> commonly used compression programs. The gzip program by
> comparison uses a history buffer of 32 kbyte and bzip2 uses a
> history buffer of 900 kbyte.
>
> Biggest disadvantage is that it isn't streamable; presumably, it has
> to work on seekable datasets. But if one replaced zlib with rzip for
> compressing the tarballs, presumably one could get similar or even
> better performance to a binary delta without losing information or
> changing the format.
At one point in time it was also 50-60 times slower than bzip...
http://olstrans.sourceforge.net/release/OLS2000-rsync/OLS2000-rsync.html
Though this article claims it is now faster than gzip:
http://www.linuxjournal.com/article/8051
So... further testing is probably needed :p
Jeremy Shaw.