Minutes of the special guest RT Conference featuring Tom Almy, Electrical Engineer at Tektronix, Inc. Contents Copyright (c) 1990, GEnie Forth RoundTable. Released for duplication and redistribution in whole or in part, provided all origination credit is included. Sub: Minutes of guest RT Conference. Guest: Tom Almy, Tektronix, Inc. Topic: 'User Interface for non-Forth Users' Date: 05/17/90 Time: 22:26EDT Attendees: [[Tom Almy] FIGGUEST] [[paul] P.THUTT] [[Gary] GARY-S] [[Len] NMORGENSTERN] [[John W] J.WAVRIK1] [[Dennis] D.RUFFER] [[Kevin] APPERT] [[IRVING] I.MONTANEZ] [[Wil] W.BADEN1] Minutes: <[paul] P.THUTT> Haven't seen you since FIG chapter went to Corvallis Any last moment questions, Tom ? <[Tom Almy] FIGGUEST> No, I'm ready Until we go formal you will see the interlacing (party-line) exchanges I discussed.. is here. is here. Dr. John and Len, welcome <[Len] NMORGENSTERN> Hi Gary, Glad you weren't flooded out. is here. 1 minute, folks ah, I see they got the name change on that account

Dennis, thanks to you and Tom for the name fix !!! 1/2 minute is here. Let's roll....

The GEnie Forth RoundTable is very pleased to welcome this ... evening's special guest, Tom Almy, an Electrical Engineer ... for Tektronix, Inc. in Beaverton, Oregon the past 17 years. ... Mr. Almy's involvement with Forth began in 1980, using MMS ... Forth on a TRS-80. In 1982 he wrote a Integrated Circuit CAD ... program in Forth on the TRS-80, which is still in use in a ... revised form on 80386 I-clones. ... There are about a dozen users of the current edition, none ... of whom are particularly computer literate. ...

Mr. Almy has also written a commercial Forth for the Z-80, ... as well as Native Code Compilers for Forth for ... Z-80, 80x86 and 80386 processors. ... His subject for tonight's GEnie Forth Guest Conference ... is 'User Interfaces for non-Forth Users', which his CAD ... program makes ample use of. ... - Please make welcome Tom Almy ... .. .. .. .. .. ga Tom is here. <[Tom Almy] FIGGUEST> I have a deep concern with keeping Forth so that the "average" person (one who is not computer literate).. can make use of the PRIDE CAD program without going nuts! or having to get ahold of me for each little problem.. <[Tom Almy] FIGGUEST> I also have taught forth classes to over 100 people.. <[Tom Almy] FIGGUEST> And have seen where they have difficulties.. <[Tom Almy] FIGGUEST> Major changes in my system, some of which are being adopted.. <[Tom Almy] FIGGUEST> No Screen files.. <[Tom Almy] FIGGUEST> Better number formats.. <[Tom Almy] FIGGUEST> "To" variables.. <[Tom Almy] FIGGUEST> All words, control strucgtures work in interpret state.. <[Tom Almy] FIGGUEST> so that one doen't need to be concdred about being in.. <[Tom Almy] FIGGUEST> a colon definition.. <[Tom Almy] FIGGUEST> graceful aborts on errors.. <[Tom Almy] FIGGUEST> easy access to host os. <[Tom Almy] FIGGUEST> The net result is that these mainly scientists with little background.. <[Tom Almy] FIGGUEST> with programming are able to write simple forth programs and issue commands without problems.. <[Wil] W.BADEN1> is here. <[Len] NMORGENSTERN> What commands are you talking about, typically? <[Tom Almy] FIGGUEST> The most obvious is the printing function .". Why should one have to use .(? .. <[Tom Almy] FIGGUEST> But I go alot further in allowing all control structures such as IF.. <[Tom Almy] FIGGUEST> BEGIN.. CASE.. by going into compile state when these words are found by the interpreter.. <[Tom Almy] FIGGUEST> I keep track of the nesting and when the outermost structure is finished then I execute the compiled code,.. <[Tom Almy] FIGGUEST> and then toss it out!.. <[Tom Almy] FIGGUEST> This operation is much like the early Forth-like system, STOIC.. <[Tom Almy] FIGGUEST> Which compiled all code into a compile buffer.. <[Tom Almy] FIGGUEST> It didn't have an interpret state, but behaved as though it did since.. <[Tom Almy] FIGGUEST> the user interface did look alot like Forth. <[John W] J.WAVRIK1> I guess the question is: when you are writing a system for a... <[John W] J.WAVRIK1> non-Forth user do you use Forth to make an interface that does not.. <[John W] J.WAVRIK1> require programming or are you trying to bring programming. to the "average user" <[Tom Almy] FIGGUEST> The users of my CAD program do write simple forth programs.. <[Tom Almy] FIGGUEST> It is true that most of their "CODE" consists of commands which are <[Tom Almy] FIGGUEST> executed once to generate a piece of mask layout.. <[Tom Almy] FIGGUEST> but they use looping structures to place arrays of objects.. <[Tom Almy] FIGGUEST> If they had to place all there loops in colon defintions, which.. <[Tom Almy] FIGGUEST> they othyerwise don't tend to need.. <[Tom Almy] FIGGUEST> then they would simply get confused.. <[Tom Almy] FIGGUEST> Don't forget, these people have trouble with BASIC or PASCAL, which to.. <[Tom Almy] FIGGUEST> us would seem ultra-simple. (Not to belittle them, their skills lie in <[Tom Almy] FIGGUEST> other areas -- they don't want to learn the ideosyncracies of a computer language. Tom, it sounds like your CAD program... uses a TTY style interface, and the Forth... interpreter (with a few fixups like you mentioned)... works pretty good. However, I was wondering... if you have ever had to get into a more... graphics oriented interface. with pull down menus and windows support? <[Tom Almy] FIGGUEST> well there is more to it than a simple TTY interface.. <[Tom Almy] FIGGUEST> There is a graphic editor for the design.. <[Tom Almy] FIGGUEST> The editor has "pop up" windows, which are easier to use, according to my user community.. <[Tom Almy] FIGGUEST> There is als a help facility. From within the graphic editor.. <[Tom Almy] FIGGUEST> you can get help on the graphic edotor commands.. <[Tom Almy] FIGGUEST> Outside the graphic editor, many commands will produce "usage".. <[Tom Almy] FIGGUEST> messages if they are not given arguments, or are given wrong arguments.. <[Tom Almy] FIGGUEST> I make use of magic numbers on the stack to make sure the more complex.. <[Tom Almy] FIGGUEST> commands have all args present and in the right order so that I can give an error message if they are not.. <[Tom Almy] FIGGUEST> Also, there is a menuing interface for those that are scared off of command <[Tom Almy] FIGGUEST> lines. When the program starts you can say "MENU" (there is a message.. <[Tom Almy] FIGGUEST> to that effect) and then you have a menu driven interface. It even has memory of the last command arguments for quick recall. <[Tom Almy] FIGGUEST> (the graphic editor menus pop up with the cursor on the last selection as well). When you use the menu interface.. <[Tom Almy] FIGGUEST> you are shown what the command line style command would be.. <[Tom Almy] FIGGUEST> so that you can learn the command line interface while you are using.. <[Tom Almy] FIGGUEST> the menus. I found out that most users start with the menus.. <[Tom Almy] FIGGUEST> but quickly drop them after learning the commands.. <[Tom Almy] FIGGUEST> Even in the graphic editor each command can also be given from a single.. <[Tom Almy] FIGGUEST> keystroke so once you learn the keystrokes you don't need the menus at all. Again most users use the keystrokes.. <[Tom Almy] FIGGUEST> I'm so busy with other things *I* can't remember the comands myself (!). and always use the menus. quick question - is this no pain Forth a commercial or p.d. product ? (it might make a contender for the edu-forth project Wil is undertaking) <[Tom Almy] FIGGUEST> Every thing I have done would be easy to put into any Forth system.. <[Tom Almy] FIGGUEST> The PRIDE system is currently built on top of LMI UR/Forth 386.. <[Tom Almy] FIGGUEST> (It needs 32 bit integers, there is no problem with space) The alterations to the number conversion (which I haven't discussed all the error handling, etc., can be made by simply providing new routines to the existing vectored routines. .. <[Tom Almy] FIGGUEST> There was no need to go back to the kernel source, which I don't have anyway.I see no reason why these things couldn't have been done in F-PC, to use an example of a pd system. I think I see how you are doing it, but correct me.. if I'm wrong.... when you go into menu mode, you use cursor control keys... to move the menu selection item?.... that is the simplest way, but I've had to play... with EXPECT when I've doen some systems... that way, I could have menus and command mode... active at the same time... Have you played with that type of approach?... also, have you ever messed with mouse support? <[Tom Almy] FIGGUEST> The menu mode is text based.. <[Tom Almy] FIGGUEST> The first character of the menu item is unique and BOLD faced.. <[Tom Almy] FIGGUEST> That key is struck. At lower levels, sometimes expect is used but it is modified so that hitting the Escape key will cause a return to the top menu. Also hitting Ctrl-Break will cause a return to the main menu no mater what is running at the time, providing of course you are using the menu system!.. <[Tom Almy] FIGGUEST> The graphic editor uses the mouse. It is the only way to go for this application. The interface code has to pick up either keystroke commands or mouse button presses.. <[Tom Almy] FIGGUEST> while of course moving the display cursor if the mouse moves.. <[Tom Almy] FIGGUEST> and updating the coordinates at the bottom of the display. <[Tom Almy] FIGGUEST> The users say that this system runs much faster than any others they.. <[Tom Almy] FIGGUEST> have used. A tribute to Forth. It can even do all of this and.. <[Tom Almy] FIGGUEST> rubber-band animate objects on a PC/XT class machine.. <[Tom Almy] FIGGUEST> (of course not with the current 386 code, but with an earlier version). <[Len] NMORGENSTERN> How much user interaction did you have while.,. <[Len] NMORGENSTERN> you were developing the program? <[Tom Almy] FIGGUEST> LOTS!.. <[Tom Almy] FIGGUEST> My office is right in the midst of the user community.. <[Tom Almy] FIGGUEST> even though my manager's office (where I would normally be near) was several hundred feet away!.. <[Tom Almy] FIGGUEST> Proximity to the user community is very important!.. <[Tom Almy] FIGGUEST> When I first started this program I was a user of it myself which is better than not knowing anything about the needs. <[Tom Almy] FIGGUEST> but I had a real user within a year.. <[Tom Almy] FIGGUEST> (We ar talking about 1982, here. This program has been.. <[Tom Almy] FIGGUEST> honed for many years, and it is still being improved). <[John W] J.WAVRIK1> ok <[John W] J.WAVRIK1> Ever try an interface in which Forth is used to write... <[John W] J.WAVRIK1> a simple problem-oriented user language? <[Tom Almy] FIGGUEST> Not a meta language, which I assume you mean (ie a language written.. <[Tom Almy] FIGGUEST> in Forth which creates (DOES>?) words that the application uses).. <[Tom Almy] FIGGUEST> But I have done some work with object oriented Forth, where Forth words <[Tom Almy] FIGGUEST> represent complicated, sometimes dynamic structures.. <[Tom Almy] FIGGUEST> These are used in Pride to represent routing cables.. <[Tom Almy] FIGGUEST> There would be no way to do this in a traditional language.. <[Tom Almy] FIGGUEST> and one would have to result to Smalltalk to get sufficient power.. <[Tom Almy] FIGGUEST> and I would't want to subject these folks to the Smalltalk programming.. <[Tom Almy] FIGGUEST> language, and it is too dificult to modify it to user needs.. <[Tom Almy] FIGGUEST> We also use Forth in some interfacing applications where words are written <[Tom Almy] FIGGUEST> specific to the task, but *everybody* does that! I suppose that would be a problem-oriented language! Dr. wavrik - you have a follow-up ? ga <[John W] J.WAVRIK1> Actually I had something simpler in mind... <[John W] J.WAVRIK1> I was impressed years ago with the ease with which MINITAB could be.. <[John W] J.WAVRIK1> used. This is a statistics package. Command lines look like.. <[John W] J.WAVRIK1> ADD column C1 to column C2 and put the result in C3 <[John W] J.WAVRIK1> looks like English... <[John W] J.WAVRIK1> The trick is that the first word is a command word.. <[John W] J.WAVRIK1> most of the other words are noise... <[John W] J.WAVRIK1> All you really need is ADD C1 C2 C3... <[John W] J.WAVRIK1> I tried this in Forth for an application and then dropped <[John W] J.WAVRIK1> it when I went on to other things.. <[John W] J.WAVRIK1> I was curious whether the idea of creating... <[John W] J.WAVRIK1> an artificial mini-language was ever explored... <[John W] J.WAVRIK1> by others in Forth who do write commercial programs.. <[John W] J.WAVRIK1> (I don't).. <[Tom Almy] FIGGUEST> The conciseness.. <[Tom Almy] FIGGUEST> of Forth is appreciated here.. <[Tom Almy] FIGGUEST> No one minds having to do arithmetic on the stack.. <[Tom Almy] FIGGUEST> after all, it is just like their (HP) calculators which are prevalent around here.. <[Tom Almy] FIGGUEST> I suppose there would be opposition when in places where.. <[Tom Almy] FIGGUEST> infix calculators are the norm.. <[Tom Almy] FIGGUEST> And these people know mathematics and don't minde concise <[Tom Almy] FIGGUEST> expressions. I don't think they would appreciate the extra typing of the English sentences.. <[Tom Almy] FIGGUEST> In the graphics editor they like the simple commands.. <[Tom Almy] FIGGUEST> there is never more than two layers of menus, and the menus alternate rather than being heirarchical.. <[Tom Almy] FIGGUEST> Other systems give them confusing sets of heirarchies that may or may.. <[Tom Almy] FIGGUEST> not match the heirarchy of their mind sets. Closing remarks please, Mr. Almy <[Tom Almy] FIGGUEST> Well.. <[Tom Almy] FIGGUEST> The time has flown, and I hardly had a chance to talk about all I have done.. <[Tom Almy] FIGGUEST> The result is that these users have no fear of using Forth.. <[Tom Almy] FIGGUEST> Managers of course still do.. <[Tom Almy] FIGGUEST> But they are happy with the results.. <[Tom Almy] FIGGUEST> Namely the designers are able to create masks for IC's, hybrids, whatever their job.. <[Tom Almy] FIGGUEST> in short time.. <[Tom Almy] FIGGUEST> using a program that runs on a relatively inexpensive system with little training. <[Tom Almy] FIGGUEST> I have found that this system is good for playing around in.. <[Tom Almy] FIGGUEST> Since the changes represent *extensions* rather than drastic.. <[Tom Almy] FIGGUEST> surgery to Forth, existing programs will run.. <[Tom Almy] FIGGUEST> and I have a more bullet proof environment,.. <[Tom Almy] FIGGUEST> convenient DOS access, etc. Tom, on behalf of the GEnie Forth RoundTable,'Thank you'... for your views on an area many Forth (and other) programmers ... tend to overlook, the program's presentation to the user.

All may stay and talk, but this conference has officially ended. === End of Steno notes. ===