In order to fit [VAX000...] and [VAX86A...] on one tape at 1600
bpi the cumulative index, INDEX.LIS, has been squeezed with LZCMP.  For
convenience copies of the squeeze program (LZCMP) and the unsqueeze program 
(LZDCM) are provided on this directory.  They were submitted by Glenn Everhart
and also appear in [VAX86B.RCAS86.CBIN].

	Here are the commands I used to squeeze INDEX.LIS:

$LZC = "$DUA7:[VAX000.INDEX]LZCMP"
$LZC -B -X 0 -M 16 -V INDEX.LIS INDEX.LZS

	Here are the corresponding commands to unsqueeze INDEX.LZS:

$LZD = "$DUA7:[VAX000.INDEX]LZDCM"
$LZD -B -X 0 -M 16 -V INDEX.LZS INDEX.LIS

	Suggestions and comments:

	1.  Do not do squeezing on a busy system.

	2.  Set WSEXTENT as large as possible.  Mine was set to 500 and
my process generated 3,387,151 page faults during the squeezing operation.

	3.  The amount of compression is impressive - 8358 blocks to 1871
blocks or 77.61%.  The performance - 1648.68 seconds cpu time and 3 million 
page faults is not.

	4.  However, decompression does not seem to be a problem - 138 seconds
cpu time and only a few page faults - with WSEXTENT set to 4100 the most
physical memory used was 738 pages.

	(If this seems rambly it is partly because I am writing it as I
carry out the operation.)

	Unsqueeze operation was successful - differences reports no
differences.

Notes from compression of Dallas index:

	o   I ran LZCMP with a larger WSEXTENT (approximately 2000 pages)
and experienced a much faster squeeze than Joe did with the previous index.
I ran the job interactively, so I don't have exact stats, but it ran in
a small fraction of the time noted above (with a peak WS size of about 1K
pages).

	o   Unsqueeze results were comparable to previous, again with
no differences detected.

- Tom Gerhard