Has there ever been an APL-ASCII transliteration scheme which achieved any significant adoption?

Viewed 298

Around the time I started working with APL in the 1980s, there was a transliteration scheme for writing APL code to file by Jim Weigang. While the scheme itself was solid, one would probably never want to program with it. One transliteration scheme I remember (just found it, APL Bang!) looked as if it could have a chance. Both Weigang's notation and APL Bang! are mentioned here:

More info on APL Bang! (which appears to have gone through a revision)

More recently, there is the ELI language which has its own transliteration scheme, only slightly similar to APL Bang! Some info on ELI is here:

There were a few other transliteration schemes from APL*Plus/PC and others, but these were things which played second fiddle to APL and one could program with but only in the case of emergency. I know of J and K and Q, but I would not think of these as "transliterations", but redesigned languages.

2 Answers

No.

I've been involved with APL for over 30 years, and while I am aware of all the ones mentioned, I've only seen a select few in real life, and then only as a last resort due to constraints of the medium.

While I obviously have not seen all or everyone's APL code, I can confidently testify that no such transliteration scheme has achieved any significant adoption.


A related development is a kind of visual transcription scheme where one "draws" APL characters using ASCII symbols, and then combines them. E.g. -> makes and xx makes ×. However, this is purely an input method, and the proper Unicode character immediately substitutes the ASCII art. It is available as a bookmarklet.

To my knowledge, the only APL systems which supplied their own built-in transliteration were STSC APL Plus/PC and the APL systems for the CDC 6600-series mainframes, APLUM and maybe the one by CDC itself. The APL Plus/PC transliteration looked something like this:

A #is B / #iota #rho B

while the APLUM one looked something like this:

A $IS B / $IO $RO B

I tried the STSC transliteration once on a PC without the required APL character generator ROM. The CDC translation was evidently designed to be used with a Model 33 Teletype terminal. (You might see one of these in a museum.) These features were built into the system and, as Adam correctly stated, used as a last resort due to the constraints of the medium. These were, to my knowledge, never widely used.

Having said that, for applications we wrote on mainframes (Sharp, STSC, VSAPL, etc.) in the mid 1970s, it was normal to put a hook into character prompting such that a programmer could get in and do some maintenance, such as drop a file component, change a variable, or copy something in. In one case it was even possible edit functions with the help of an APL alternative to the DEL editor, APL's version of EDLIN. These would be backed with our own transliteration function for use on non-APL terminals, which implemented something like the APLUM scheme (as it was the only one around in the mid 1970s) and was copied wherever and whenever needed. Miserable to use but a lifesaver when needed. In summary, in the early days, transliteration was there, you just had to implement it yourself.

Some applications on timesharing (such as Sharp's 39 MAGIC) were designed to be usable on non-APL terminals. When signing on to a timesharing service, the system "knew" what sort of terminal you were using and would support it accordingly. You could get reasonably far without APL characters. Further, many of the early terminals such as the IBM 2741 and various AJ and Xerox Diablo daisywheel printers, had interchangable type elements, such that the worst possible problem was a keyboard without APL keytops. It was only with some of the cheaper ASCII terminals where lack of APL support was an issue. ASCII transliteration often wasn't perceived as necessary.

I suppose users who really needed ASCII transliteration moved on to J.

Related