ATN stuff

[FTP:DEWOLF at WPAFB-AFAL]

Hi Dick,

I've been trying to get this message to you all afternoon, but
WPAFB-AFAL has been down.  If you are desperate for this stuff,
(and need info or help), give me a call at home (217-344-6169).
I don't expect to be at CSL or on the machine next until approx.
2:00pm on monday.

,
$s
Hello Dick,

Firstly, let me tell you that we found the FASLOAD bug in
LISP 1305.  It appears that it is in the code that you and
Quux wrote this August.  Apparently, when FASLOAD tries to 
get the last group of 6 blocks, the monitor raises an 
END OF FILE condition (but still returns the last few blocks).
LISP notices this but doesn't check to see if any blocks were
returned.  Anyway, it works when we changed it to load one
block at a time.

You are welcome to have my ATN stuff.  All together there is about
230 blocks (128 words/block) of code.  I'm not sure how much you
want, so I'll list all the relevant files here.
The format is:

	file name	size in 128 word blocks		description


These files are in [1000,130,source].  The comprise the ATN system
(interpreter, compiler, support functions, etc.)

	atn.lsp		50	atn interpreter
	lex.lsp		22	morphology routines, etc.
	parse.lsp	8	top-level function to apply an ATN
	loaddic.lsp	10	dictionary manager
	hatn.lsp	10	helpful auxilliary functions
	atncompiler.lsp	30	compiler for ATNs
	simplify.lsp	16	optimizer for ATN compiler

These files are in [1000,130,atn].  They comprise a hacked up (by me)
version of the Woods parser (syntax only!).

	s.net		6	  |
	np.net		13	  |  ATN
	vp.net		9	  |  networks
	random.net	8	  |
	extra.atn	8	auxilliary functions called from nets
	simple.dic	3	mini-dictionary
	parse.ini	2	init file for parser

These files are on [5,720,source].  The contain macros needed 
by above files.

	define.lsp	28	define macro
	lmacro.lsp	11	lisp macros

The .FASL versions of the .LSP files can be found in [1000,130]
for the first group and [5,720] for the last.  

How do I get them to you?  I see three basic options:

   (1)	Put them on a mag-tape and slip in the mail.
	(is this fast enough?)
   (2)	Try to get them to WPAFB and then over the ARPA net.
	One problem is that we have only a 1200 baud
	line to WPAFB.  The CONNECT program slows it down
	to boot.  A more serious problem is that CONNECT
	tends to hang very often is either machine is loaded
	(i.e. when the sun is up).  If CONNECT is used, then,
	it should be at night.
	   Also, we can't use FTP from WPAFB (no
	network privledges there).  I don't know if you can
	run FTP from SAIL and suck up files from WPAFB (you
	might give it a try as an experiment).  If not, we
	could always use MAIL to send the files from WPAFB
	to SAIL.  If you type a â??G to the MAIL program at
	WPAFB it will ask you for a file name and include
	its contents in the body of the messsage.

   (3)	I could try to write a UNIX compatable tape here 
	with our PDP11 (which can run UNIX) and take it
	to CAC, put it on thier machine and use FTP to
	get it to SAIL.

How to use CONNECT to transfer files?  The documentation
is in DOC:CONNECT.DOC (on our machine and probably also
on the one at WPAFB).  The idea is that you get CONNECT
running on both machines, tell one:
	RECEIVE <filename>
	START
and tell the other:
	SEND <filename>
	START 
Depending on where you are relative to the direction of 
transfer, you have to do these in the proper order and
may have to type a character to get it going.

If you want to use CONNECT, I'll try to send more coherent
instructions.


Let me know what you want to do.  I'm happy to help.

				- Tim

p.s.  I told the MIT INQUIRE program to have any network mail
sent to me at MIT-?? forwarded to DEWOLF@WPAFB-AFAL.  Thus,
if its any easier, you can mail to FININ@MIT-AI (or MC, etc.).
I usually check WPAFB-AFAL for mail every weekday.





// 30 //