[fitsbits] Rice compression from the command line

[Mark Calabretta]

On Wed 2006/07/19 09:47:45 -0400, Arnold Rots wrote
in a message to: Mark Calabretta <mcalabre at atnf.csiro.au>
and copied to: Rob Seaman <seaman at noao.edu>, fitsbits at nrao.edu

>I did a simple test using bzip2 for event files on Solaris.
>Sure, I got 15% savings over gzip and I'm not worried about the
>compression taking about twice as long.  But the decompression taking
>an order of magnitude longer than gunzip bothers me.  Even if it is
>more than compensated for by what's gained in transmission time, the
>users' perception will be that things are worse, since decompression

I agree that for data distribution it's the end-user experience that
counts.

However, I'm skeptical of the order of magnitude figure.  The worst I
saw was x3 in elapsed time which is what a user would perceive but
typically it was between x1.5 and x2.  Do you have figures for this
test similar to those I sent?

Cheers, Mark