Unicode

The advent of Unicode solved many problems with dealing with APL characters, however there was still some wiggle room as to which Unicode codepoint were to be used in a Unicode implementation of APL, and different implementors made different choices. This article, which documents these differences, is adapted from an original paper by Bob Smith that attempted to raise awareness of these issues because the differences impede transfer of information.

The relevant document for the APL character set is the APL Character Repertoire (ACR). For whatever reasons, that document never became a standard, but it does provide some guidance, and is better than each implementor making separate choices.

Introduction

There are a surprising number of similar APL characters in Unicode and in several cases some implementors went one way, others the other way. The following table lists the characters in question, along with the way APL2, Dyalog, GNU APL, NARS2000, ngn/apl, and dzaima/APL behave. APL2000 states that Generally the default codepoint scheme for the VisualAPL product follows the IBM APL2 workstation scheme. Please edit this page if you believe there are other characters that should be included in the table.

When there are differences among APL implementations, users can become confused. They type something into one APL system, copy it to another and are greeted by a SYNTAX ERROR or the like.

The whole basis for the confusion in a lengthy thread on comp.lang.apl entitled caret vs and is that in some implementations the symbol for the logical And function is U+005E only, in some implementations it's U+2227 only, and in some both characters work. The original poster encountered some APL text from the APL Wiki that had been produced by a system that supports U+005E and copied it into a system that uses U+2227 only and fails on U+005E.

When our systems differ in the set of acceptable characters for the same function, it serves only to confuse the end user to the detriment of the community.

Comparison of implementations

APL name	Glyph	Codepoint	Unicode name	APL2	Dyalog APL	GNU APL	NARS2000	ngn/apl	dzaima/APL	Monadic	Dyadic
Epsilon	∈	U+2208	Element of	No	No	Yes	Yes	No	No	Enlist/Type	Membership
	∊	U+220A	Small Element of	Yes	Yes	Yes	Yes	Yes	Yes
Minus	-	U+002D	Hyphen-minus	Yes	Yes	Yes	Yes	Yes	Yes	Negate	Minus
	−	U+2212	Minus sign	No	No	Yes	Yes	Yes	No
main content awaiting approval
Circle	○	U+25CB	White circle	Yes	Yes	Yes	Yes	Yes	Yes	Pi Times	Circular
	⚪	U+26AA	Medium white circle	No	No	Yes	Yes	No	No
Diamond	⋄	U+22C4	Diamond operator	Yes	Yes	Yes	Yes	Yes	Yes	Statement Separator
	◇	U+25C7	White Diamond	No	No	No	Yes	No	No
	◊	U+25CA	Lozenge	No	No	Yes	Yes	No	No
	⬦	U+2B26	Diamond	No	No	Yes	Yes	No	No

These characters are included here because they have been encountered them in APL code displayed somewhere on the Internet or in a PDF file. Thus blindly copying them into an APL session can produce an error which might well confuse the user.

Functionality

The following statements can be used to test the functionality of the symbols:

⍎⎕← '1',(⎕UCS 16⊥2  2  0 10),'1'   ⍝ Epsilon
⍎⎕← '1',(⎕UCS 16⊥2  2  0  8),'1'   ⍝ Epsilon
⍎⎕← '1',(⎕UCS 16⊥2  2  1  2),'1'   ⍝ Minus
⍎⎕← '1',(⎕UCS 16⊥0  0  2 13),'1'   ⍝ Minus
⍎⎕← '1',(⎕UCS 16⊥2  2  2  3),'1'   ⍝ Modulus
⍎⎕← '1',(⎕UCS 16⊥0  0  7 12),'1'   ⍝ Modulus
⍎⎕← '1',(⎕UCS 16⊥2  2 12  6),'1'   ⍝ Star
⍎⎕← '1',(⎕UCS 16⊥0  0  2 10),'1'   ⍝ Star
⍎⎕←     (⎕UCS 16⊥2  2  3 12),'1'   ⍝ Tilde
⍎⎕←     (⎕UCS 16⊥0  0  7 14),'1'   ⍝ Tilde
⍎⎕← '1',(⎕UCS 16⊥2  3  7  1),'1'   ⍝ Nor
⍎⎕← '1',(⎕UCS 16⊥2  2 11 13),'1'   ⍝ Nor
⍎⎕← '1',(⎕UCS 16⊥2  3  7  2),'1'   ⍝ Nand
⍎⎕← '1',(⎕UCS 16⊥2  2 11 12),'1'   ⍝ Nand
⍎⎕← '1',(⎕UCS 16⊥2  2  2  7),'1'   ⍝ And
⍎⎕← '1',(⎕UCS 16⊥0  0  5 14),'1'   ⍝ And
⍎⎕← '1',(⎕UCS 16⊥2  2  6  4),'1'   ⍝ Not More
⍎⎕← '1',(⎕UCS 16⊥2 10  7 13),'1'   ⍝ Not More
⍎⎕← '1',(⎕UCS 16⊥2  2  6  5),'1'   ⍝ Not Less
⍎⎕← '1',(⎕UCS 16⊥2 10  7 14),'1'   ⍝ Not less
⍎⎕← '1',(⎕UCS 16⊥2  5 14  6),'.=1' ⍝ Jot
⍎⎕← '1',(⎕UCS 16⊥2  2  1  8),'.=1' ⍝ Jot
⍎⎕← '1',(⎕UCS 16⊥2  6 10 10),'1'   ⍝ Circle
⍎⎕← '1',(⎕UCS 16⊥2  5 12 11),'1'   ⍝ Circle
⍎⎕← '1',(⎕UCS 16⊥2 11  2  6),'1'   ⍝ Diamond
⍎⎕← '1',(⎕UCS 16⊥2  2 12  4),'1'   ⍝ Diamond
⍎⎕← '1',(⎕UCS 16⊥2  5 12  7),'1'   ⍝ Diamond
⍎⎕← '1',(⎕UCS 16⊥2  5 12 10),'1'   ⍝ Diamond
⍎⎕←     (⎕UCS 16⊥2  5 10 15),'←1'  ⍝ Quad
⍎⎕←     (⎕UCS 16⊥2  3  9  5),'←1'  ⍝ Quad
⍎⎕←'1{',(⎕UCS 16⊥2  3  7 10),'}1'  ⍝ Alpha
⍎⎕←'1{',(⎕UCS 16⊥0  3 11  1),'}1'  ⍝ Alpha
⍎⎕← '{',(⎕UCS 16⊥2  3  7  5),'}1'  ⍝ Omega
⍎⎕← '{',(⎕UCS 16⊥0  3 12  9),'}1'  ⍝ Omega

Note that the last four lines will not work on a system that doesn’t support dfns.

Atomic Vector

If the Atomic vector (⎕AV) has no room in which to include these new characters, an implementation can translate them on entry to the corresponding symbol that is in ⎕AV. NARS2000 even has a means of translating symbols on the way out via Copy (Ctrl+C in Windows) to various other APL systems that don't support the same set of principal characters NARS2000 uses for the functions in the above table.

Considerations

Unicode was a great start to enabling APL characters to be used, however in order for there to be interoperability, implementors have to agree upon which characters are functional. It doesn't matter if one's system can change the mapping of glyphs to codepoints as the vast majority of users won't change from the default behavior. Implementors therefore have to decide if it is worthwhile to support the above codepoints.