Roger, I made remarks along those lines many moons ago, about the power of adverbs and conjunctions, so please don't feel you have to preach to me about such things. This is from my APL93 paper, "The Role of APL and J in High-Performance Computation", but I'm relatively sure I have said such things much earlier than that:
"A semantically rich language is of immense value because knowledge in one area benefits another -- learning ten verbs and twenty adverbs gives the potential for specifying 200 different actions. Similarly, a computer language with conjunctions and adverbs offers richness of expression to the programmer." What I am trying to point out is that,as you noted, there is no need to change J. However, if a front end, or skin, can improve the lot of those who are either unwilling to learn new notations, or who are more comfortable with a different presentation of it, why not make it easy for them to create an interface that pleases them? It might even attract new J users. As for the Babel (or Babylon, or perhaps Babble On) paragraph, there was some humor intended there. Please don't take it seriously. Bob Roger Hui wrote: > Fine. You are welcome to immerse yourself with character > design, and I am happy to stick with language design. > You did interject yourself into a discussion on introducing > special characters into J and what I responding to is that > interjection. > > >> This way, mathematicians could use "sigma" (I can't >> find the glyph on my keyboard...) for +/, English >> language bigots could use "sum", and certain other >> types could use "AddUpTheElementsOfTheArgumentArray", >> depending on their respective whims. I recommend that this >> skinnability be dubbed "Babel". >> > > Here's the thing. Even if sigma majuscule were a symbol on > the keyboard, to introduce that is to detract from one of > the main tenets of J (and APL), namely adverbs (operators). > Once I understand what +/ does, I would have a pretty good > understanding of what */, times/, or/, and/, max/ etc. do. > But if what I know is sum or sigma or > AddUpTheElementsOfTheArgumentArray, how do I know > how to multiply the items of an array? How am I supposed > to know that */ is PI, or that +/ and */ are strongly related? > > Adverbs are very efficient. If you know m verb symbols > and n adverb symbols, you know how to do m*n things. > If you know m verb symbols and n "sum" or "sigma" like > symbols, you just know how to do m+n things. > > > > ----- Original Message ----- > From: Robert Bernecky <[email protected]> > Date: Thursday, March 12, 2009 12:12 > Subject: Re: [Jgeneral] J with APL > To: General forum <[email protected]> > Cc: 'Geoff Streeter' <[email protected]>, Adrian Smith > <[email protected]>, Alexander Skomorokhov <[email protected]> > > >> Hi, Roger. >> >> It sounds like you're letting the tool (IE) drive design, rather >> than the other way around. A few comments: >> >> 1. Cut/paste: In Linux, you merely highlight the desired text, >> with a mouse click or swipe, then do >> the middle-button >> click to paste. No need for any typing >> (Ctrl-c). I admit that >> this is a small potatoes issue. >> >> 2. IE is on the way out, according to a few recent erag articles. >> So, perhaps all your problems with it will go >> away, when time >> gets full. >> >> 3. I was not proposing any changes to J in my message. >> What I was saying was that the problems of >> entering, displaying (and printing) unicode >> characters, including APL, >> are finally over. Well, at least in Linux... >> >> 4. What might fit in nicely is a user-defined "skin" for J, >> acting as a >> session >> manager in lieu of the J session manager, >> much as various media >> players allow a user to customize their >> player with software that >> presents the player's user interface in a >> different way. >> >> This would leave the existing J engine in its >> current form, yet let >> users >> write their scripts in any format they >> choose, and display the results >> in any format they desire. Yet, the formal >> interface to the engine >> would remain unchanged (I think), and J >> scripts could be exchanged >> as always in a uniform manner. >> >> This way, mathematicians could use "sigma" (I >> can't find the glyph >> on my >> keyboard...) for +/, English >> language bigots could use "sum", and >> certain other >> types could use >> "AddUpTheElementsOfTheArgumentArray", depending on >> their >> respective whims. I recommend that this >> skinnability be dubbed "Babel". >> >> Rockless Bob >> >> Roger Hui wrote: >> >>>> This is an iota; Alt-i: ? >>>> This is an omega; Alt-w: ? >>>> This is a transpose; Alt-Shift-6: ? >>>> >>>> >>> So here I am in Windows, where the equivalents are: >>> iota; ctrl-i >>> omega; ctrl-w >>> transpose; ctrl-shift-6 >>> >>> Fine. So now I want to copy some text in my Dyalog >>> >> session. >> >>> I highlight the text and my fingers, without any intervention from >>> my brain, hit ctrl-c. Oops. (ctrl-c is the set >>> >> intersection> symbol, for you non-APL-aficionados.) >> >>> The main point I want to make is that J has 7-bit >>> ASCII characters and you'd have to have rocks in >>> your head to introduce special characters into J. >>> The following quotes from >>> http://aplteam2.com/aplwiki/BrowserTest >>> are instructive: >>> >>> All versions of Internet Explorer get it wrong. >>> >>> Many browser under Windows are capable of >>> displaying a wide range of APL symbols. However, >>> at least some symbols are missed. Favorites are >>> ⌿⍀⍝⍎⍕⍪ and the symbols for NAND and NOR. >>> >>> I am not interested in discussions about character set >>> issues and how it's a solved problem, etc. Dyalog has >>> indeed done a magnificent job (miraculous, even) with >>> the APL characters. But: the bottom line is that J >>> does not have a character set problem and let's keep >>> it that way. >>> >>> >>> >>> ----- Original Message ----- >>> From: Robert Bernecky <[email protected]> >>> Date: Thursday, March 12, 2009 8:19 >>> Subject: Re: [Jgeneral] J with APL >>> To: General forum <[email protected]>, Morten Kromberg >>> >> <[email protected]>, 'Geoff Streeter' <[email protected]>, >> Alexander Skomorokhov <[email protected]>, Adrian Smith >> <[email protected]>> >> >>> >>> >>>> I'm finally fairly happy with APL text and LaTeX, vi, >>>> web browser, and APL windows now, thanks >>>> to the following: >>>> >>>> - Geoff Streeter (and probably others) at >>>> Dyalog, They have created a Unicode-based >>>> APL system including standard mods to >>>> >> the XFree86 >> >>>> (Linux windowing) configuration that >>>> >> let me enter >> >>>> APL characters directly into the APL >>>> >> session,>> using keystrokes like these. >> If you don't >> >>>> like the >>>> keystrokes, you can customize them to >>>> >> your >> >>>> own whim: >>>> >>>> This is an iota; Alt-i: ? >>>> This is an omega; Alt-w: ? >>>> This is a transpose; Alt-Shift-6: ? >>>> >>>> What I find entirely delightful about >>>> >> this >> >>>> approach is that >>>> the keyboard changes also work >>>> >> EXACTLY the >> >>>> same way in vi, >>>> web browsers, etc. I can >>>> >> copy/cut/paste APL >> >>>> text across >>>> apps with no problems. >>>> >>>> - Adrian Smith, for his APL385 font. >>>> >>>> - Alexander Skomorokhov, who told me about >>>> >> xelatex and >> >>>> its ability to crank out APL >>>> >> with no fuss. >> >>>> The above red text was directly copy/pasted from a pdf file >>>> produced with xelatex, from this text, which I tinkered from slides >>>> for a talk I'm giving at NYU on March 24. The red text is what >>>> makes the APL happen: >>>> >>>> \documentclass[handout]{beamer} >>>> \usepackage{beamerthemesplit} >>>> >>>> \usepackage{fontspec} >>>> \setmainfont{Arial} >>>> \setmonofont{APL385} >>>> >>>> \title{Tacit Parallelism} >>>> \author{Robert Bernecky} >>>> \institute{Snake Island Research Inc} >>>> \date{\today} >>>> \usecolortheme{default} >>>> >>>> \begin{document} >>>> This is an iota; Alt-i: {\tt ?} >>>> >>>> This is an omega; Alt-w: {\tt ?} >>>> >>>> This is a transpose; Alt-Shift-6: {\tt ?} >>>> \end{document} >>>> >>>> I'm using Linux, but Dyalog also has a BillWare(tm) >>>> version of their unicode system. >>>> >>>> I'm not totally happy with Adrian's font, as I don't think >>>> it has the balance (if that's the right word) and elegance of >>>> Joey Tuttle's APL fonts, and his font seems to pick >>>> typefaces more or less at random. E.g., the dollar sign in >>>> the following looks OK here, but in the pdf file, it is >>>> clearly out of place with the remainder of the text: >>>> >>>> x =. 2 3 $ 4 >>>> >>>> My thanks to all for making life much simpler. >>>> >>>> Bob >>>> >>>> Don Watson wrote: >>>> >>>> >>>>> Raul >>>>> >>>>> The non-ASCII characters were a bit of a mess. This was a >>>>> >>>>> >>>> first attempt at >>>> >>>> >>>>> something and I am not used to creating such script. I hand >>>>> >>>>> >>>> wrote the >>>> >>>> >>>>> non-ASCII characters onto the printed page and scanned the >>>>> >>>>> >>>> result into a PDF >>>> >>>> >>>>> file. >>>>> >>>>> J programs would still be transferable, because there is no >>>>> >>>>> >>>> change to >>>> >>>> >>>>> storage format - the two ASCII characters are still stored. >>>>> >>>>> >>>> The change is >>>> >>>> >>>>> only when those characters are output. >>>>> >>>>> Don >>>>> >>>>> ----- Original Message ----- >>>>> From: "Raul Miller" <[email protected]> >>>>> To: "General forum" <[email protected]> >>>>> Sent: Thursday, March 12, 2009 2:20 AM >>>>> Subject: Re: [Jgeneral] J with APL >>>>> >>>>> >>>>> On Wed, Mar 11, 2009 at 11:15 PM, Don Watson >>>>> >>>>> >>>> <[email protected]> wrote: >>>> >>>> >>>>> >>>>> >>>>> >>>>>> http://bcompanion.com/Compromisepdf.pdf >>>>>> >>>>>> >>>>>> >>>>> Is there a reason you put this in a pdf ? I ask, >>>>> >> because >> >>>>> >>>>> >>>> I recently >>>> >>>> >>>>> uninstalled adobe's reader because of a security flaw. >>>>> >>>>> http://www.adobe.com/support/security/advisories/apsa09-01.html >>>>> >>>>> I am installing the new version of adobe now, but if you needed >>>>> to use pdf because regular text was inadequate for your proposal, >>>>> I think that says something about the proposal itself -- it would >>>>> mean -- for example -- that you are proposing that J >>>>> >>>>> >>>> programs should >>>> >>>> >>>>> not be transmittable using regular email. >>>>> >>>>> ... >>>>> >>>>> Ok, I have downloaded the proposal now. The hard part, >>>>> >>>>> >>>> obviously,> is the characters themselves. (Personally, >>>> >> I >> >>>> could read the >>>> >>>> >>>>> J code but had problems understanding the "compromise >>>>> equivalent verb". I am not sure if my problems were because >>>>> of the ways the intended characters were rendered or whether >>>>> my problems in understanding were because I was >>>>> >> expected to >> >>>>> understand some character(s) which I did not understand.) >>>>> >>>>> > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
