Hi Bowerbird,
Your compression result on KJB
4,430,861 bytes (original size)
is intriguing.
1,413,207 bytes (Stuffit alone)
958,211 bytes (your best recipe)
OK. However you should be using a state-of-the-art
compression tool like bzip. What figures does bzip
give? Try MacBZip at:
persephone.cps.unizar.es/~spd/bzip2
and repeat your test.
As I have said before bzip uses some of the ideas
you are using, and even recursively.
MacBZip alone gives about 995,000 bytes.
Now you have already done 4% better. OK. Does
MacBZip much improve your 958,211 figure?
There's another possible disappointment: the batch
size of MacBZip is only about a Meg. So, the real
test is to compare with a BZip compression that
compresses in a single batch. Now anyone with a
linux machine can hopefully get a king size BZip to
do this more stringent test. Anyone?
I'm really root'n for you. The ordering of strings
may have some new magic in it.
Cheers
larry s