Next: Project Correspondence Up: index.html Previous: Acknowledgements

Project Log

At the outset of the sniffle project, I began to record each day's work in a log file. This file, listed below, provides a fairly complete account of the evolution of the project. It also contains lots of information about product vendors, ``gotchas'' that I encountered during the development process, and anything I considered worth writing down.


Log of activity on remote control via Web project with Prof. Littman.

Author:    Terence Kelly  <tpkelly@phoenix.princeton.edu>
RCS File:  $Source: /u/tpkelly/littman/RCS/log,v $
	   $Revision: 1.9 $

6/6/95	Meeting with Jay Lieske and Kirk Alexander in ICGL
	They recommend hardware encoding of MPEG.  Mention IRIS "InPerson"
	software for SGI machines.  Names of CIT people who have clue:
	Kevin Perry (Mathematica instructor), Dave Harrington (CU-SeeMe
	experience).  Names of other people with clue:  Ken Stieglitz,
	Bede Liu (image processing professor).

	Littman discuss funding with Diane Balestri.

	Suggest net search of "CU-SeeMe", "Live Video", "MPEG". See
	Netscape page for server push/client pull.

	Basic idea that emerged was to have two parallel systems solve
	different parts of the problem:  Use Web to transmit initial
	conditions from client to lab-based server and to start
	experiment.  Meanwhile a separate system handles video feedback
	from lab to client end using perhaps CU-SeeMe.

6/7/95	Met with Littman in lab.  He approves of idea of two parallel
	systems for instrument operation and video feedback.  I said
	I'd perform appropriate research.

6/8/95	Wrote Ted Nadeau asking for MPEG etc. advice
	Faxed and snail-mailed to many many companies from magazine of
		Kirk Alexander requesting MPEG hardware info etc.

	Set up mail-logging scheme.  Mail littman instead of mlittman
	to log mail to ~tpkelly/littman/mbox.littman file.

	Brief bobw on HTML and LaTeX

	Surf net looking for httpd server info.  Looks like Netscape
	requires Windows NT.  No mention of Linux among flavors of
	Unix compatible with Netscape server.

	NCSA mentions Intel-based PC at URL
	http://hoohoo.ncsa.uiuc.edu/docs/setup/PreCompiled.html

	Windows and Windows NT httpd at URL
	http://www.yahoo.com/Computers/World_Wide_Web/HTTP/Servers/
	  Windows___Windows_NT_HTTPD/

	Cool Windows (not NT or 95, just Windows) server URL:
	http://www.city.net/win-httpd/

	Wrote to rdenny@netcom.com about above server, asking minimum
	hardware requirements.

6/9	Got reply from rdenny, who says "A 386 should be fine, as long
	as it has 4MB or more of memory."

	CGI-DOS, helps you use Perl with Win-httpd, URL:
	http://www.achilles.net/~john/cgi-dos/

	Download Bob Denny's Windows httpd .zip file.  Note the
	directions on installation (we need recent PKUNZIP):

	"Create a directory c:\httpd (don't be creative yet!),
	and unzip the kit, using the PKUNZIP "-d" option to
	preserve directories. Then open the file
	C:\httpd\htdocs\index.html with your browser and read
	it.  Once you have gone through the checklist, start
	the server and access it as http://1.2.3.4/ where
	1.2.3.4 is your PC's IP address. Try out the demo
	page. If the server exits immediately, look in
	c:\httpd\httpd.log for the reason."

	Found CU-SeeMe page:  http://cu-seeme.cornell.edu/

	Sunday June 11 Dartmouth will broadcast a CU-SeeMe video of
	Bill Clinton's commencement speech:
	HTTP://picard.dartmouth.edu/~oly/BillVision.html
	There's a reflector at U-Penn, IP address 130.91.72.36,
	according to Dartmouth docs.

	CU-SeeMe Windows system requirements:
	http://cu-seeme.cornell.edu/supportedwin.html
	(looks like 386DX will suffice, but cheapest video board
	will cost around $350)

	Remote control on the Internet:
	http://www.indstate.edu/msattler/sci-tech/comp/
		misc/remote-control.html

	CU-SeeMe User's Guide:
	http://www.indstate.edu/msattler/sci-tech/comp/CU-SeeMe/how-to.html

6/10	Install CU-SeeMe on OPR PC in anticipation of Clinton Speech.
	Connected to U-Penn reflector and caught glimpse of video broadcast.
	IT WORKS!

	MBONE FAQ at AT&T:  http://www.research.att.com/mbone-faq.html

6/11	Watched Clinton speech at Dartmouth.  Very disappointing.
	At best got 1 frame per second.  Average was around 0.5 fps.
	Worst was 0.1 fps.  Don't know source of bottleneck.

6/14	Talked to Littman.  He found interesting new Web page on remote
	controlled CU-SeeMe camera:
	http://www.hiof.no/ludvigsen/webdoc/pt/pt_dummy.html

	He said I should start to configure httpd server on 386 on
	cart, not to touch Robert's 386 unless other one is short of
	disk space.  Our 386s have microchannel architecture, not
	EISA.  The one on the cart has a phone-jack type ethernet
	connector, Xoftware and Novell Winsock (?).  It's a PS/2
	Model 80.

	Phone numbers:  G-101, where "George" often is: 5567
			Littman's office:  5198

	FTP'd Windows HTTPD and pkunzip.exe to C:\httpd\ on 386
	on metal cart.  That machine is a nightmare.  Slow, poorly
	connected, software not configured properly (Xoftware is
	particularly uncooperative).  Tried to pkunzip -d per the
	instructions but ran out of disk space.  May have to do this
	on Robert's PC.

6/18	On 6/17 or thereabouts installed WinHTTPD on D:\ partition of 386
	machine on cart in G-101.  Configuration was a hassle because the
	httpd assumes you're installing on C:\, not D:\.  But it works
	and successfully served a "hello world" document for over 24 hours.
	The URL is http://douglas.princeton.edu/.

6/19	Installed Netscape on two more PCs in G-101 (previously installed
	it on 386 on cart).  Installed VBRUN300.DLL in C:\WINDOWS\SYSTEM
	on 386 on cart.  Tested server:  all features other than imagemap
	now seem to work properly.  CGI-DOS and Visual Basic CGI stuff and
	Windows CGI stuff all seem to work.  Had to edit PIFs for some DOS
	stuff to make it all work and make error messages go away.  Server
	is still up and running as of 10:53 a.m.

6/24	Got "hello world" program to compile on Robert's PC and emit
	digital output to DVM via RTI-205 board and STB-50 connector
	cable.  Program toggled voltage across outputs between 5.04 VDC
	and 0.033 VDC.

	Problem 1: Could not compile a hello program on PC on metal cart
	where server is because compiler complained about DECL.TPU
	unit, even though I copied it directly from RTI205 directory,
	and then from Robert's C:\TP directory, where it worked.  I
	think the problem may be that metal cart PC has later version
	of Turbo Pascal (a downgrade).

	Was able to compile a HELLO2.EXE on Robert's PC and then run
	it on my PC, but it didn't produce the proper range of output
	voltages:  output voltage oscillated reliably between 4.23 VDC
	and 4.37 VDC at correct times.  Problem with STB-50? RTI-205?

7/6	Met Littman at noon and discussed plan for future.  Here is current
	plan:

		    1.	Client submits amplitude parameter to
			server, which passes parameter to batch
			file or one program therein.

		    2.	O-scope begins to read & record data.

		    3.	Pascal program gets amplitude parameter
			passed by client, outputs that voltage.

		    4.	Everybody waits for a while.

		    5.	Data is read from O-scope to file on server,
			then returned to client in native .WFM format.

		    6.	Data is displayed at client using 222UTIL\VIEW.EXE

	Sal Desiano gave me numbers for Analog Devices company (makers
	of RTI-205 board):

			Main		617-461-4111
			Regional	215-643-7790
			Tech. Support	800-262-5643
			Sales		617-461-4194

	At noon demo'd to Littman activation of "Hello, World" program
	on server PC from Netscape client (the program output an analog
	voltage via the RTI-205).

	In evening played with Tektronix 222 O-scope.  Idea:  use Tektronix
	utility VIEW.EXE (in C:\222UTIL) to view data captured by scope at
	client end:  server transmits a .WFM ("waveform") file to client
	with new MIME type e.g. "application/waveform", and client spawns
	local copy of VIEW.EXE to view it.

7/8	Yesterday and today tested idea of using VIEW.EXE as Netscape
	helper application -- it works from my PC at OPR.  It causes
	Netscape to issue a "floating point error: stack under" message
	on G-101 PS/2, but this is obviously not a problem with the basic
	idea.

	Bill Vinje told me about "Lab View" -- a GUI for lab equipment
	interfaced to PCs, and about a standard protocol for communications
	beteween lab instruments and PCs called ??? (I forget) which allows
	a PC equipped with a special board to talk to a daisy-chain/bus of
	o-scopes, etc..  Must ask Littman about this.

	Must call Tektronix and ask them for latest PC, Mac and X versions
	of VIEW utility.

7/9	Got following scheme to work well with RC circuit that operates on
	human time scales (using roughly 10 meg resistor and .22 uf cap):

	Client submits voltage, delay1, delay2.  Server activates Pascal
	program that drops output voltage to zero, waits delay1, raises
	output to voltage specified by user, waits delay2 sec., drops
	voltage to zero.

	Can't get batch file that invokes above Pascal program to follow
	it with a GETWFM.EXE call -- resulting waveform file is empty.
	Possible solution:  have Pascal program talk to oscilloscope to
	get waveform, rather than GETWFM.EXE.

7/10	Met with Littman in G-101 to demo above.

7/11	Looked at Tektronix ftp site ftp://ftp.tek.com/mbd/support
	where lots of utilities for 222 scope live.  Unfortunately none
	of them are newer than the ones we have.

	We should perhaps use TOADIF utility to convert .WFM files to
	.ADF format at server end, and view with WAVIEW at client end.
	May be difficult in any case to get client-spawned viewer to
	display two scope traces easily.

	Problem: disk space is running low in lab.

7/17	Lately have had problems with RS-232 interface to Tektronix 222.
	Want to write little program to send command-line arguments to
	scope and write whatever comes back to standard output.  This
	program CAN send strings like "BUT 1" to scope and scope usually
	responds appropriately.  But a run-time hardware error (usually
	code 162) occurs when attempt to read from scope.

	Called Tektronix customer support at 1-800-835-9433 x2400 (got
	the number from WWW page), spoke with Dave Frazel at x3028.  He
	said problem might be our firmware version:  we have version 2.16
	whereas latest version is 3.8.  He said he'd have one of his
	software people call me later today.  The Tektronix customer support
	line is open 6:00 am to 5:00 pm Pacific Time.  Tektronix customer
	support e-mail:  tm_app_sup@tek.com.

	Spoke with Ed Lingell about problem with 222.  He wants to know what
	the error code means.  He said Tektronix doesn't have source to
	222UTIL stuff.  Fellow who wrote it left company.

	Later spoke with Tektronix Dave Nelson, who suggested that a
	"handshake error" might be responsible -- handshaking is set on the
	RS-232 board on the PC, in part using the DOS mode command.  The
	settings that Nelson uses with Windows terminal are:  CRLF, No parity,
	8 data bits, 1 stop bit, no parity check.  Nelson also gave me the
	number of Cal Diller, who wrote the 222UTIL stuff, and is now working
	for "Metro Tek":  503-640-4906.

	Firmware upgrade:  Washington DC service center (800-962-9738) says
	they don't handle 222 scopes.  Referred me to Beaverton OR factory
	service (800-835-9433 x2400).  Also gave me Ship To address:

			Tektronix, Inc.
			Attn:  Factory Service, M/S 78-576
			Regional Distribution Center
			Beaverton  OR   97077

	Still waiting for pricing info on firmware upgrade from Tektronix.

	Spoke with Barry Long of Analog Devices, requested info on all
	RTI-205-like products.  He said recent versions of drivers are
	available from BBS:  617-461-4361, 8 data bits, 1 stop bit, no
	parity, 1200-9600 baud.  Barry Long is an application engineer,
	his number is 617-461-3090.  Application support is 617-461-4111.

	Spoke with Cal Diller, author of 222UTIL stuff.  His number at
	Metro Tek is given above, his e-mail is 71662.31@compuserve.com.
	He is VERY helpful.  Said he'd mail me the source to 222UTIL stuff.
	Said in order to handle COM port well would have to write interrupt
	handler to deal with interrupt that occurs whenever the SINGLE BYTE
	of COM port buffer is written.  222UTIL things do NOT have an
	interrupt handler, just continuously read from port in tight loop.
	Said there is a Pascal command to get a character from the COM port.
	Said Visual Basic includes routines that do interrupt handling.
	Visual Basic is the way to go, says he.  Recommends book, "The C
	Programmer's Guide to Serial Communications," by Campbell, published
	by SAMS.  This book contains assembly code hardware interrupt
	handlers.

	Can also write Windows DLLs in C/Pascal to use Windows interrupt
	handlers, and these can be called from Visual Basic.

	Diller said he could give me source to API he wrote that lets you
	talk to instrument via hi-level commands, it runs as DLL and requires
	Visual Basic variable-length strings.  Visual Basic costs $299 at
	Programmer's Paradise; Borland Delphi is $99, similar to Pascal.

	Finally, Diller recommends National Instruments and their Lab View
	product:  6504 Bridge Point Parkway, Austin TX 78730-5039,
	512-794-0100.

7/19	Installed Visual Basic on PC AT with 486 upgrade & 4 megs RAM.
	Have not tested yet.  Wrote Cal Diller to ask if he knows about
	reading COM1 using Visual Basic.  Documentation is silent about
	this.

7/21	Ordered MS Visual Basic Professional Edition thru CIT MDC,
	Nina Moyer 8-6798.  Will deliver tomorrow (Saturday) to Littman
	home.  Charged to account 150-1009.  MDC orders from a company
	called "Ingram Micro" in or near Buffalo NY.  Ingram Micro sales
	is 800-456-8000, customer support is 800-274-4800.  They must
	receive an order by 3 p.m. to ship that day.  MDC is 8-5566, open
	M-F 10:30-4:30 (at least at this time of year).

	Egghead software in Lawrence was not helpful, could not get us
	product with educational discount.  MDC price is $90, Egghead is
	$340.

	Spoke with Wally Schwanke of Tektronix, 503-627-6630 regarding
	firmware upgrades.  He said Tektronix now has 5-day tunaround time
	from time of receipt of scope to time of return shipment.  He will
	investigate cost of upgrade and call back with info.

	Nina Moyer of Microcomputer Distribution Center placed order of
	MS Visual Basic Pro, should be delivered to Littman home tomorrow.

7/23	Installed Visual Basic Pro, tested MSCOMM.VBX control.  It works
	well.

7/24	Cal Diller's package arrived containing a diskette.  Diller's
	post-it note on diskette says "I think the functions you want are
	in LIB\ASYNCXX.C".  I copied the files on the diskette to
	C:\TPK\DILLER\ on my PC at work.

	Also in Diller's package is a description of product by Diller's
	company, Metratek (503-693-1607):  Waveform Manager Pro ($199).
	Will talk to 222 scopes, write GIF files.  But can it be made to
	run in batch mode?  Maybe using Visual Basic!  Will ask Diller
	when I write to thank him.

7/25	Talked to three third-party vendors who make software for RTI
	series boards.  DSP Development dude had little clue, said he'd
	send literature.  HEM Data Corp sells boards, fully supports the
	815 board, will send literature.  LABTECH dude Brian Harrington
	was clueful & helpful, told me about LABTECH product Notebook (Pro)
	for data acquisition & talking to boards.  Has some sort of macro
	feature called "batch runner" which sounds like exactly what we're
	looking for.  Will send literature on products including educational
	discounts and bulk discounts.  Phone numbers:

		DSP Development		617-577-1133

		HEM Data Corp.		810-559-5607

		LABTECH			508-657-5400

	I found out about these companies in the Analog Devices catalog.

	Spoke with dude at DSP (toll free 800-424-3131 x400) who explained
	that their current products are more oriented toward data analysis
	rather than acquisition, therefore not appropriate for us.

7/27	Wally Schwanke of Tektronix (503-627-6630) said we could possibly
	get firmware upgrade for free but would require more details on
	how we got the scopes.  Litman said Tektronix approached him around
	1987 and offered to donate fancy equipment, he wrote up proposal,
	Tektronix gave us a dozen 222 scopes and one color printer.  He
	was disappointed by what they gave us, wanted more.  Gift came from
	"Signal and Measurement Division" or "Test and Measurement Division."
	Gift was in support of science education.  Littman said could obtain
	copy of his proposal from Mary Lavin of MAE (melavin@pucc, 8-5127)
	or Anna Faiola.  I wrote Mary Lavin at PUCC to ask for proposal.

	GREAT IDEA:  For dealing with all RTI board issues, simply write
	an INTERPRETER in Turbo Pascal or whatever that takes a text
	stream of commands and makes the RTI board do its thing based on
	those commands.  E.g. if it gets a line that says "DOT 161" it
	tells the RTI board to output binary 161 on its eight digital
	output wires!  Presto!  We then have a LANGUAGE that can be
	transmitted over the Web that causes the RTI board to do ANYTHING
	it can possibly do!!  Damn I'm good.

7/28	Got Visual Basic program with no forms whatsoever to run and
	respond; it returned something to the client.  The thing to do
	is to have NO FORMS WHATSOEVER in your project.  Use ONLY code
	modules, per R. Denny's suggestion in WinHTTPD documentation.
	Have a subroutine "Main" in one of the modules and designate this
	as the first thing to happen upon program execution.

	Asked Littman about software & manuals for RTI-815.  He authorized
	me to buy $120 worth of them from Analog Devices.

8/4	Purchased documentation, software and screw terminal board with
	ribbon cable for RTI-815.  Tested board with EXER utility and
	DVM, found that analog outputs work.  Modified Visual Basic
	sample program, tested it, found that it works.  We get two
	analog outputs and possibly eight digital outputs.  Have not
	yet been able to get digital outputs to work.

8/5	Digital outputs don't work because they operate off a SECOND
	set of output pins from a 34-pin J1 jumper in the middle of the
	RTI-815 board.  Must purchase this ribbon because we need
	digital outputs.

	One place to start looking for the GIF format spec is at
	  http://www.cis.ohio-state.edu/hypertext/faq/usenet/graphics/
	  fileformats-faq/part3/faq.html

	Here's what that document has to say about the GIF format:

	GIF is a data stream-oriented file format used to define the
	transmission protocol of LZW-encoded bitmap data. GIF images may be up
	to eight bits (256 colors) in depth and are always compressed. Despite
	the fact that GIF supports only 8-bits worth of colors, and the
	multimedia extensions introduced in the 89a release have not been
	widely utilized, GIF still remains a popular choice for storing lower
	resolution image data.

	The GIF89a specification is available via many BBSs and on-line
	information services. You may also obtain the specification directly
	from CompuServe:

	  CompuServe Incorporated
	  Attn: Graphics Technology Department
	  5000 Arlington Center Boulevard
	  Columbus, OH 43220
	  Voice: 614.457.8600, 800.848.8199

	Lots of GIF utilities, plus a .ZIP file containing the format,
	are at http://ubu.hahnemann.edu/SimTel/gif.html.

	The format spec itself is at
	http://documents.cfar.umd.edu/imageproc/gif89a.doc

8/8	Ordered two proper 34-pin female J1 plugs (one with, one without
	polarizer) and ribbon cable from Master's Electronics in Raritan,
	908-725-3533.  Will be shipped today via first-class mail, arrive
	later in week, charged to my credit card.  Cost will be under $15.

	Fellow at Master's Electronics says that for LED hex readout of
	TTL lines, need 9368 chip (around $7 each) and 7-segment hex LED.
	He will see if he has pinout documentation on 9368 chip.  Will take
	at least 2 weeks to order.

	Littman hooked up his own hex LED displays for me, so no need to
	purchase.  Tested digital outputs and they work well.  Pinouts on
	hex LEDs are worth recording, because they're unusual:

       	       	       	   ________
		  Vcc   1  |  _   | 14 	 Vcc
		 bit 1  2  | |_|  | 13  bit 2
		 bit 0  3  | |_|  | 12  bit 3
			4  |      | 11
		  GND   5  |	  | 10
		        6  |	  |  9
		  GND   7  |______|  8 	 GND

8/10	Ordered from Free Software foundation Windows gnuplot diskette
	for Littman project; Texinfo manual for me; DJGPP CD-ROM and DOS
	utils diskettes for OPR.  All should arrive via FedEx tomorrow.

	Littman says re: Analog Devices bill to call purchasing, get PON,
	then have Analog Devices re-bill me.  Call "Helen" in MAE if
	problems; she works in a.m. only.  We didn't discuss FSF software
	(gnuplot for Windows) or $10 for J1 parts.

	Littman and I added COM3 and COM4 ports to 486 PC; will use COM3
	for Tektronix scope.  Also replaced homebrew J1 plug with proper
	plug.  Tested both scope (using 222UTILS) and RTI-815 digital
	outputs (using EXER), both seem to work.

	At this point we don't know where our video server will live.
	Littman promises a video board before he leaves town on Saturday
	(Aug. 12 -- 2 days from now).  We don't know what PC it will live
	on.

8/15	Interpreter basically works.  Transitions on analog outputs require
	about 5 milliseconds.

8/17	Tested long "programs" containing digital outputs on my interpreter
	yesterday.  When you submit a commandstring consisting of very many
	"DOT 0; DOT 1" commands, the time required for these commands gets
	longer as the commandstring "executes."  The interval for which the
	low bit remains at 0 or 5 V begins at around 50 milliseconds, if
	I recall correctly, and increases up to around 250 milliseconds.
	Finally the VB program crashes with an "Out of string space" error
	message.

	This led me to develop an un-locker to remove the lockfile in cases
	where the WEBRTI.EXE crashes before removing it at normal termination.

	Today got an imagemap working.  You see an 8-arrow'd Pan-n-Tilt
	controller form.  Click on e.g. the down arrow and you get back
	an HTML doc saying "You hit DOWN."  There's a bug in the R. Denny
	docs on the subject:  you must enter "imagemap.exe" in the URL
	to which the ISMAP image is submitted, not just "imagemap" as the
	documents say.

8/19	Met with Littman for a while, demo'd the CU-SeeMe camera controller
	and WEBRTI program.  He suggested that I add ability to read in
	digital and analog values from RTI-815.  I told him I can do this.
	We went over stepper motors, he gave me a two-motor pan-and-tilt
	apparatus and documentation on how to use a chip to step the motors.
	He also showed me how to use an open-collector inverter to shield
	the RTI-815 from the deadly 12 V logic involved with the motor
	controllers.

	Later in the evening I cleaned up the WEBRTI interpreter a lot.
	Now it gives correct and nice diagnostic feedback, correctly
	tracks "recursion" (i.e. "SOURCE" commands), etc..

	Next order of business:  Implement commands like "TEK," "LOOP,"
	"DOTB," and input commands like "DINB," "DIN," "XAIN."

8/22	The fellow at CIT who co-ordinates the Fall Information Fair
	is Jon Edwards, jedwards@phoenix.  Maybe Littman and I can
	demo there.

	Late at night spent much time getting Tektronix scope to talk to
	Visual basic.  Crux of problem was proper settings of serial port
	memory addresses and IRQ numbers.  The final working settings are
	therefore worth recording:

				COM1		COM2

		Address:	03F8		02F8
		Baud rate:	2400		2400
		Parity:		N		N
		Data bits:	8		8
		Stop bits:	1		1
		IRQ:		4		3

	It took an obscene amount of effort to get these settings right.

8/24	Spent a long time on the Tektronix routines.  The behavior of
	the routines is probabilistic.  When I send button commands to
	the scope to freeze the screen and stash waveforms CH1 and CH2
	in RAM (with a view toward retrieving it shortly thereafter;
	the idea is to freeze both waveforms at the same time), sometimes
	things work correctly, other times the waveforms don't get stored.
	Might have to slow the routine down way slow to make it work,
	but early tests of this approach did not work well.

8/26	Thought of a great name for the project:  SNIFFLE, the Simple
	Network Interface to the Full Functionality of Lab Equipment.
	(Thought of name "sniffle" days ago; it's the name of a friend's
	Saab sedan; only later realized acronym fit.)

	I think that as of today the functionality of the Visual Basic
	program at the heart of SNIFFLE is complete and error-free.
	Now I must test the system thoroughly, work out all bugs, and
	document it.

	Also on the to-do list:  wire up the stepper motors and play
	with pan-n-tilt implementation.

8/28	Called AIRPAX company, manufacturer of stepper motors, regarding
	the motors we're using for pan-n-tilt stuff.  They say our model
	K82231 is obsolete, but they'll FAX me complete specs on it.
	Last night I had trouble wiring up SAA 1027 controller chips because
	I lacked info on which resistor values were appropriate to our
	motors.

	AIRPAX says their distributor is Sager Electronics, 800-724-37800
	(yes, that phone number is correct -- just dial all of those digits
	and all will be well).  Sager, like AIRPAX, won't give out prices
	on products until you're ready to buy.  Weird.

9/5	Check to see if Visual Basic has an "eval" command.  This would
	let us send VB programs from client to server and let us replace
	our toy macro language with a true programming language containing
	powerful commands.  I think I investigate this question months ago,
	but Littman suggested the idea today and I said I'd check it out.

9/6	Got pan-n-tilt imagemap control panel to work.  Must be careful with
	GET-method URLs in .map file:  if they're too long problems arise.
	length limit seems ridiculously short.  Probably due to fact that
	URL args are *last* thing passed to cgi program on command line.

	Cleaned up source code by shortening all lines to under 75 characters
	(except for one very long line that Visual Basic won't let us break).

9/7	Littman found a CU-SeeMe reflector on campus:  mongo.princeton.edu.
	Littman says that X-terms can view the broadcast from our CU-SeeMe
	server via MBONE something-or-other.  We haven't gotten this to work
	yet.

	Upon investigation we discovered a product called "nv" which
	supposedly displays CU-SeeMe on an X-terminal.  It is installed in
	/usr/princeton/bin/conferencing/, along with other related utilities
	for internet teleconferencing (e.g. audio & whiteboards).  The nv
	man pages are not installed by CIT, so it's best to get them from
	the source:  ftp://parcftp.xerox.com/pub/net-research/nv*.  I've
	put the man page in my ~tpkelly/man/man1 directory, along with a
	.ps paper on the subject of nv.

	Littman reports a bug in session logs of SNIFFLE:  Sometimes when
	a barrage of commands is issued, or when two clients try to use
	the server, bad things happen, e.g. wrong session log is returned.
	I confirmed this problem when trying to demonstrate locking to
	him.  Dunno what's up until I stare at the code.

	Fixed bug in session logs later at night.  Now every time the
	WEBRTI program executes a different log is generated.  When too
	many log files pile up in a special session log directory, the
	client is warned about it and given an opportunity to delete
	them with a single click.

9/14	Documentation complete.  This log to be included in my project
	report as an appendix.


tpkelly@cs.CS.Princeton.EDU
Thu Sep 14 02:35:48 EDT 1995