The APL Orchard

12:04 PM

@Adám So when I multiply an integer by 2 in APL, does it do <<1 (left bit shift by 1) under the hood?

@KritixiLithos Probably, but it has to check whether to upgrade the datatype (and for overflow). I'll ask.

I'm trying to tackle something and I am taking in efficiency into account, so that's why I'm asking

@KritixiLithos Why don't you compare the execution time for 2× and 3×?

2× is slightly faster

12:19 PM

@KritixiLithos Also, it may depend on what data you throw at it, and how much. You'll have to play around, or wait for my colleague to respond.

(I'm not working today, so I sent him an email.)

thanks

12:51 PM

> We will contact the top finalists in April 2017.

What?!

@Zacharý Where?

The email with the title "Congratulations on finishing the Dyalog APL Challenge 2017" from dyalog_apl@sqore.com

@Zacharý Grrr. Those Sqore people!

And (it's okay if you're not allowed to answer this) how does one prove student status?

@Zacharý Sqore takes care of that part. IIRC, they wait until we chose (potential) winners, and then contact those directly, asking them to prove it in whichever way.

12:57 PM

Okay.
The distance between (1 1) and (3 4) is 5. LOL.

what do you mean?

@Zacharý No, that would be between (0 0) and (3 4).

I know, just making fun of a mistake in the student competition (They fixed it now)

which question is it?

I'm not sure if I can say that ...

1:02 PM

@Zacharý We fixed it? Did you complain?

No. I didn't complain about it, it got fixed before I noticed it.

I'm pretty sure it's not from this year

@KritixiLithos No, but it will in 17.0. Similarly for multiplication by 3, 4, 5, 7, 8, and 9.

wait how does it do that for numbers like 3?

I'm assuming APL will never have bitwise operations?

@KritixiLithos It won't, it will just multiply by 3 probably.

1:05 PM

@KritixiLithos *2+1, i.e. shift left 1, perform addition

what

5*3 != 5*2+1

n+n×2

@KritixiLithos leave the APL notation aside

@Zacharý No. APL philosophy dictates that a number is a number. Also, how would you shift bits on DECFs?

ah k

1:07 PM

DECF's?

@Zacharý Decimal Floats.

help.dyalog.com/16.0/Content/Language/Introduction/…

Bitshifts are easily done with 2n and floor(n/2).

@Adám you shift the relevant part of the mantisa, or change the exponent

mantissa?

@KritixiLithos mantissaEexponent

1:12 PM

so that's the base

@KritixiLithos No, the base is 10 (implied in the E)

@KritixiLithos no, the mantissa part is multiplied by 10**exponent

@Uriel Wouldn't that give a different result?

could you give an example, I'm still not entirely sure I understand what it is

@KritixiLithos 1.234E56 The mantissa is 1.234

Significand

The significand (also mantissa or coefficient) is part of a number in scientific notation or a floating-point number, consisting of its significant digits. Depending on the interpretation of the exponent, the significand may represent an integer or a fraction. The word mantissa seems to have been introduced by Arthur Burks in 1946 writing for the Institute for Advanced Study at Princeton, although this use of the word is discouraged by the IEEE floating-point standard committee as well as some professionals such as the creator of the standard William Kahan. == Example == The number 123.45 can be...

1:16 PM

@Adám ignore the exponent. But the mantissa shifted by 2 would multiply the whole floating point by 2, so no

@Uriel Depends on which way you shift.

Of course certain precision is to be considered

@Zacharý huh? I know two ways to shift: right, and left

How sould shifting the mantissa right multiply by 2?

@Zacharý You can always swap to data representation (⎕DR), perform your transformation, and swap back.

@Zacharý left, not right. right is for division

1:19 PM

When did you specify this was left?

I never specified left or right. I just mentioned shifting the mantissa

1:34 PM

@Zacharý I'm curious to whether you've ever actually needed bitwise operations other than on PPCG.

Oh, I feel so stupid now. Convert to binary. Apply operation. Convert back to decimal.

@Zacharý Is that sarcasm?

Sort of. I can't believe I didn't notice that before.

@Zacharý Note what?

The fact that you would be able to do one of ∨∧~≠ to the binary form(s) of the argument(s) in order to achieve |&~^.

1:44 PM

@Zacharý Oh. So then it wasn't sarcasm. In 17.0 we should have the ⍢ Under operator, and if we use the unassigned monadic ⊤ as equivalent to the current 2⊥⍣¯1, then you can write f⍢⊤ to do a bitwise f.

2⊥∧/2 2 2⊤5 7 for 5&7.

Seems like the under operator is going to be very powerful.

@Zacharý You can already play with it now: Under←{⍺←⊢ ⋄ ⍵⍵⍣¯1 (⍵⍵ ⍺) ⍺⍺ (⍵⍵ ⍵)}

@Zacharý Yes, especially since there are all kinds of cool inverses defined (but undocumented), like 'abc'∘,⍣¯1 will strip a leading 'abc' (and error if there isn't any)!

Why aren't they documented? Is it for the same reason as Screen Map?

@Zacharý There is also a proposed ⍫ Obverse: f⍫g derives a new function identical to f, but where f⍣¯1 is defined as g.

@Zacharý No, they are just "obvious". No? The inverse of prepending something is stripping it from the head.

Oh. The 'abc' gets composed into , then the inverse is applied.

1:53 PM

@Zacharý exactly. But you don't have to compose.

@Zacharý Btw, Dyalog's inversion system penetrates complex definitions, i.e. any function derived from invertible functions is invertible.

I can't imagine what an APL dialect MADE for code-golfing would look like (probably not much different, to be honest)

@Zacharý You can also invert scans, like +\⍣¯1 and ,\⍣¯1.

@Zacharý Jelly?

@Adám That's more J for codegolfing with a custom codepage

@Zacharý J is pretty much a superset of APL, less all the ⎕ utils.

J's commands are practically a superset of APL. J is not a superset of APL (at least in the way Objective-C is a superset of C)

2:01 PM

@Zacharý What is the difference? J is mostly just spelled differently (and treats 2-trains differently). One exception is ⍳ on a vector, where I find APL's definition way more powerful and useful than J's.

@Adám wasn't Roger a co-creator of J?

J already has had Under and Obverse for years. And so too with almost all the new primitives in 16.0. Only exception is dyadic ⊆ which isn't really new, as it was there under the name ⎕ML←3 ⋄ ⊂.

And you can't forget on MAJOR change in J.
indexing.

@Uriel Yes, and he is gradually adding J primitives to Dyalog APL. We also have Marshall now, another J developer. J development is pretty dead by now: Roger is teaching his son APL, not J!

J will always live on in code-golf.

2:05 PM

@Zacharý True. Its removal makes for a cleaner language, but darn, bracket indexing is convenient! Try expressing Matrix[2 4;6 8] in J!

Heck, Fortress even uses [] instead of _ for indexing/subscripting.

@Zacharý K went the other way, and embraced bracket indexing as the main syntactic feature.

Link to usages?

@Zacharý Usages of what? Brackets in K?

Yes.

2:08 PM

@Zacharý k.txt but be warned, in K everything is golfed – even the documentation. That "page" is pretty much the entire documentation.

@Adám well, my generation would probably be the last one to write code anyway

The next would just transmit their brainwaves

@Uriel I doubt it.

The generation that will be the last to write code is the one that creates gAI (general AI) capable of writing code.

@Zacharý We've long considered adding ⎕ESP…

Sarcasm?

2:13 PM

@Zacharý Well, I mean, it would be nice. The only reason for not doing so is that our C programmers are not good enough to implement it.

What would ⎕ESP do???

@Zacharý It takes any input and returns the result that the user wants.

Extrasensory perception

Extrasensory perception, ESP or Esper, also called sixth sense or second sight, includes reception of information not gained through the recognized physical senses but sensed with the mind. The term was adopted by Duke University psychologist J. B. Rhine to denote psychic abilities such as intuition, telepathy, psychometry, clairaudience, and clairvoyance, and their trans-temporal operation as precognition or retrocognition. Parapsychology is the study of paranormal psychic phenomena, including ESP. Parapsychology has been criticized for continuing investigation despite being unable to provide...

So, are the C programmers the butt of the majority of jokes at Dyalog?

@Adám then you'd recreate bubblegum

@Uriel How so?

@Zacharý I'm the only APL programmer at our HQ.

2:17 PM

@Adám bubblegum is officially described as "does it's best to guess the developer intention"