On Tue, 19 Mar 1996, Phil Knaack wrote:

> Actually, it would seem to me that a "magic bit sequence" could occur
> in most any data stream, be it compressed or not.

It can, and does.

> But for any given piece
> of data, the probabilities of it existing in the raw data versus the
> probabilities of it existing in the compressed version should be almost
> completely disjoint.

I'd call that a good theory if it weren't for the fact that I've been bit 
by this both with compressed and uncompressed sup runs.  The probability 
is affected by both the size of the data set and the size of the known 
"magic bit pattern" (which is something like 8 or 12 bytes in a very 
regular pattern).

> What I'm trying to say is that perhaps for a given amount of sup data,
> one may work and the other may not. If the one you try first fails, try
> it the other way.

Yes, I can agree with that.  I'm just trying to point out that one 
shouldn't _expect_ one or the other to work, that's all.  It tends to 
increase the level of frustration. :-/

--scott