The Forth Column

Filling out the "ANSI ASC X3/X3J14 Technical Proposal" form

Page: Please number each page of your proposal

Title: A short phrase that characterizes your proposal on this subject, please list their titles and dates.

Related Proposals: If you have submitted other proposals on this subject, please list their titles and dates.

Proposal(): Are you proposing a specific change? Then Check the Proposal box.

Comment(): Otherwise, check the comment box.

Keyword(s): Pick a keyword that helps others understand what area of Forth your proposal addresses. For example, Kernal, double integers, system word set...

Forth words(s): List all affected Forth words, including ones added or deleted by your proposal or discussed in your comments.

Abstract: Briefly convey the nature of your proposal (or comment).

Proposal: State your proposal in specific terms. When proposing a new word, or changes to an existing word, simply state the new definition. Include a stack picture if appropriate.

Discussion: If you are making a comment, put it here. If you are submitting a proposal, provide compelling arguments in favor of your proposal. BE concise. If you are aware of arguments against your proposal, state them and rebut them. Is your proposal consistent with current Forth conventions? Does the proposal involve hardware or other system dependencies? Does it affect application portability? Does it have implications for multi-user environments? Should the proposed change be mandatory? Does it affect ROMability? Is it general purpose? How does it affect execution speed? How does it affect compilation speed? What are its memory requirements?

Submitted by: Provide your name, address, adn daytime phone number.

Use the "ANSI ASC X3/X3J14 Technical Proposal Form, cont's" form if additional pages are needed. Remember to put page numbers and repeat the "Submitted by" and "Date" fields on each page.

Each member of the ANSI Forth committee will prepare a position paragraph on each issue. One member, the "magnet," will collect all the paragraphs and synthesize them into one paper describing the different sides of the issue. You can follow or participate in the discussion on any issue by signing onto the GENie Forth bulletin board, Category 10, Forth Standards. I'm told that several hundred people have already perused this category.

You will find the new Dr. Dobb's Toolbook of Forth, Volume II, to be an excellent anthology of papers on Forth programming techniques. Inside, you can find Bresenham's algorithm, FFTs, spreadsheets, and Logo, all implemented in Forth. Some papers are taken from FORML and Rochester proceedings, and others first appeared here in DDJ. Unfortunately, credit is not given to the original publication. There are 32 quality papers in all, many of them with source code. The Toolbook sells for $29.95 and is available from DDJ ([800] 533-4372 or [800] 356-2002 in California). If you don't want to type in the programs, you can get them on a disk for an additional $25.

Dick Pountain's Object-Oriented Forth (Academic Press, 1987; $19.95) begins: "Forth is, regrettably, one of the best kept secrets in the computing world." In this book, Dick extends Forth to include named objects, which increases its readability and reusability. In essence, an object is a defining word used to create defined words called instances. An instance has its own hidden data structure, made of instance variables, and a set of permitted operations, called methods. Instance variables are to instances as user variables are to tasks.

When an object is defined, all instance variable and method names are removed from the Forth dictionary by relinking the headers. Only the object itself can find them because it has the initial link, or "key," to the hidden vocabulary. All instances of this object are state smart and know the key. During compilation an instance looks the following method up in its private dictionary and compiles a reference to it.

As an example, consider complex numbers as objects. They might have instance variables for the real and imaginary parts and methods for addition, multiplication, and so on. Pountain limits his methods to @ and !, however, so that complex results are left where they belong, on the stack:

TYPE> COMPLEX
    2 VAR REAL
    2 VAR IMAG
OPS>
    :COM! IMAG ! REAL!;
    :COM@ REAL@ IMAG@;
ENDTYPE> COMPLEX

COMPLEX X 2 3 X COM!

Object-Oriented Forth (OOF?) takes you through this and other abstract data types, including lists and heaps. This book is not for beginners, and I highly recommend it.

Starting in the October 26, 1987, issue of Info World, Steve Gibson ran three articles in his Tech Talk column describing the implementation of Microsoft's QuickBASIC 4.0. According to Mr. Gibson, "The technology is known as `Threaded PseudoCode' ... and was first seen in the strange powerful language Forth." He continues, "Forth is the language you want to pretend doesn't exist because it's horrible to write with but hauntingly powerful."

Needless to say, these comments drew immediate rebuttal from the Forth community. Here is a sample, translated from Timothy Huang's Chinese Forth column: "... the term Threaded PseudoCode was used repeatedly with such an envious attitude, like a human being who just discovered the warmth of the sun. In fact, we have been using this technique for more than ten years. So why doesn't Mr. Gibson and Microsoft, as well as the whole computer industry, come straight out and recognize FORTH, and quit behaving like a dead duck with a very tough beak?"