Talk:JOSS

Just so you know: 'JOSS In' is JOSS version I, sub-version n.

disambig
I've removed this diambiguation notice from JOSS:


 * This article is on the programming language. See also religion in China and incense, Joss Stone for the British female soul singer, or Joss Whedon for the television writer/producer

I can't see the connection with religion in China and incense (neither use the word Joss), and I don't believe the personalities were known only by their first name.

Pengo 23:23, 10 August 2005 (UTC)


 * I came to the Talk page to say the exact same thing. I'm removing it until someone explains how they're relevant. - DNewhall (talk) 21:45, 3 February 2009 (UTC)

A Few Comments on Joss -- by Alan Kay
A suggestion for this article would be to not use any modern conventions, but to follow Cliff Shaw's original Joss typography and UI as closely as possible. For example, Joss was done on a very nice converted IBM typewriter (made by Tom Ellis and others at RAND), typed on fan-fold 8.5x11 paper that was drilled for one's looseleaf notebook, and the user's input was typed in green, with its replies typed in black and indented.

Another suggestion is to actually execute the Joss programs. It is annoying to have something nice like the Wikipedia, have a reasonable article about an early interactive language (like the ones on Joss and Logo) and not be able to try out examples even though we are reading the articles on a running computer that could execute them. JavaScript is more than capable for such examples, and I and a few friends decided to do a browser-only Logo embedded in a wiki to demonstrate how this could and should be. Go to LogoWiki (Don't! It's now a Russian site of some sort nothing to do with Logo) to try this. As of April 21st 2006 it is only a week old, but it should be pretty stable by the end of May.

Let's hope that the community will gradually make interpreters for the great programming languages and environments of the past!


 * That would require that the specifications for those languages and interpreters be made available on-line to the community rather than buried in the RAND library. But I don't see the point -- just as we have far better environments than typewriters with green and black ribbons typing on fanfold paper, we have far better languages than JOSS and Logo, some with on-line interpreters, e.g., http://plaid.isri.cmu.edu:8080/plaidWebTerminal/plaid.cgi -- 98.108.210.171 (talk) 00:07, 30 October 2010 (UTC)

Reference document number correction
Under References, the RAND report number of the "JOSS Users' Reference Manual" is R-1555/9, not F-1535/9. After confirming the number with the RAND library, I made this correction. A. Rizzo (talk) 15:24, 23 April 2012 (UTC)

FOCAL, MUMPS variants of JOSS, or vice-versa?
The JOHNNIAC page states, "JOSS was an ancestor of DEC's FOCAL and of MUMPS," whereas this page states, "Some [variants of JOSS] remained very similar to the original: TELCOMP, FOCAL, CAL, CITRAN, ISIS, PIL/I, JEAN (ICT 1900 series), AID (PDP-10); while others, such as MUMPS, developed in distinctive directions." So is JOSS the ancestor or the descendant? — Preceding unsigned comment added by 99.190.134.148 (talk) 13:24, 26 September 2016 (UTC)
 * Ancestor.
 * (AID-10's sources are so idiosyncratic that they might be translations from JOHNNIAC assembler). AHMartin (talk) 03:25, 10 July 2019 (UTC)


 * JOSS predates all of these. The earliest appears to be TELCOMP, which appeared after the guys from BBN visited Rand. Maury Markowitz (talk) 18:35, 1 October 2019 (UTC)

Actually, AID may be a JOSS-II, the PDP-6 version of JOSS, variant based on the source code. I’ve been poring over it for awhile now.

Also, early PDP-10 documentation mentions that it’s explicitly JOSS. “AID (for Algebraic Interpretive Dialogue) is a PDP-10 adaptation of language elements of JOSS, the well-known computing service by The RAND Corporation under contract to the United States Air Force.”(pg. 4-7 of the 1970 edition of the “PDP-10 Timesharing Handbook”, available on Bitsavers) Pashpaw (talk) 02:51, 2 October 2021 (UTC)

Floating Point reference
Article states:


 * JOSS stored all numbers as an integer and a decimal exponent. This means calculations were exact decimal values, as opposed to floating point calculations. $1⁄3$ + $1⁄3$ + $1⁄3$ was exactly 1.

I never used JOSS but I am a computer science professional:


 * 1) Floating point is an integer + exponent regardless of radix/base, it does not have to be base 2 (which prevails today). "Integer + decimal" exponent can thus be viewed as "floating (decimal) point".
 * 2) Floating point neither causes nor addresses the $1⁄3$ + $1⁄3$ + $1⁄3$ issue. Think of a base 3 system, for example. $1⁄3$ cannot be represented exactly as an integer + exponent in either binary or decimal. Having this calculation result in 1 means that something else was at play such as rounding or expression optimization (skipping the division of 1 by 3 entirely, dividing 3 by 3 instead), if "integer + decimal exponent" was used for storage. Alternatively, maybe it was a rational value (fraction, integer numerator + integer denominator) + exponent? That is another way to yield exact results in this case without rounding.
 * 3) Saying that "calculations were exact decimal values" neither helps nor can be true. Calculations are processes, not values. Their results can be exact decimal values, of course, but they will be so even if the calculations themselves are inexact. For example, a calculation may yield $1⁄3$=33·10-2. 33·10-2 is an exact decimal value, 0.33, even though it isn't exactly equal to $1⁄3$. As "10" in "base 10" has "2" as a factor (10=2·5), a any exact base 2, base 5 (or base 10) number (floating point or otherwise) can be represented exactly as a base 10 number. It is the inverse that isn't true: some base 10 numbers may not be exactly representable in base 2 or base 5.

See … nothing
“See the section on Form below” says the article. And I see, I see, I see … — Preceding unsigned comment added by 84.166.43.135 (talk) 13:24, 11 March 2024 (UTC)