The APL Orchard

7:24 AM

@brgal That isn't an official format until version 18.2, but this seems to work:

⎕FIX(⊂':namespace'),(⊃¯2↓⎕nget'cnn.apl'1),'⎕←main 0' ':endnamespace'

2 hours later…

9:05 AM

I have lost touch with iteration in Dyalog. Here is a simple Eratosthenes’ Sieve.
```apl
sieve4←{p←1+⍵⍳1 ⋄ (⍺,p) (⍵∧(≢⍵)⍴((p-1)⍴1),0)}
1 sieve4 0,20⍴1
1 3 0 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 1 0
```
It returns a pair: the arguments for the next iteration. Questions:

1. How to iterate N times?
2. How to iterate until the result stops changing?

f⍣n to repeat a function n times, and ⍣≡ to repeat a function until the return value doesn't change

Ha! Tx. I used the q symbol for Match (~) by mistake! Simple enough to make the function monadic, but are there answers for the dyadic function?

(there's no markdown in multiline messages. You only get to code-block the whole thing with ctrl+k)

you can do ⊃f/ to call a function dyadically by a list of the arguments

(would need to be (⊃{…}/)⍣)

9:22 AM

Tx! The equivalent iteration syntax in q gives me a choice between returning (a) a list of results from each iteration (c.f. Scan in APL) and (b) just the last result. An internal assignment to Quad would emulate (a) for debugging. But is there a syntax that returns (a) as a result?

9:35 AM

there is none

you have to do it manually

@SJT https://aplcart.info?q=⍣ results

@Adám could do with "cumulative" (and maybe "scan") as keywords

@dzaima Building…

10:15 AM

Now my sieve works..

      ⎕IO
0
      sieve←{⍵∧.=0:⍺ ⍵ ⋄ p←1+⍵⍳1 ⋄ (⍺,p)(⍵∧(≢⍵)⍴((p-1)⍴1),0)}
      ]disp (⊃sieve/) 1 (0,19⍴1)
┌→──┬───────────────────────────────────────┐
│1 2│0 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0│
└~─→┴~─────────────────────────────────────→┘

And I can iterate until all the 1s have been tested.

      ]disp (⊃{⍵∧.=0:⍺ ⍵ ⋄ p←1+⍵⍳1 ⋄ ⎕←(⍺,p)(⍵∧(≢⍵)⍴((p-1)⍴1),0)}/)⍣≡ ⊢ 1 (0,19⍴1)
 1 2  0 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0
 1 2 3  0 0 0 0 1 0 1 0 0 0 1 0 1 0 0 0 1 0 1 0
 1 2 3 5  0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 0 1 0 1 0
 1 2 3 5 7  0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 1 0 1 0
 1 2 3 5 7 11  0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 1 0
 1 2 3 5 7 11 13  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0
 1 2 3 5 7 11 13 17  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0
 1 2 3 5 7 11 13 17 19  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

But of course I do not need to do this. Once I have found a prime ≤ the square root of N, where N is the length of the bit vector, all the remaining 1s mark primes.

I can replace the internal test for all 0s with an external test: have we found a large enough prime yet?

      ]disp {⍺,1+⍵/⍳⍴⍵}/(⊃{p←1+⍵⍳1 ⋄ ⎕←(⍺,p)(⍵∧(≢⍵)⍴((p-1)⍴1),0)}/)⍣{5≤⊃⌽⊃⍺} ⊢ 1 (0,19⍴1)
 1 2  0 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0
 1 2 3  0 0 0 0 1 0 1 0 0 0 1 0 1 0 0 0 1 0 1 0
 1 2 3 5  0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 0 1 0 1 0
┌─────────────────────┐
│1 2 3 5 7 11 13 17 19│
└~───────────────────→┘

Now I have a scope issue. The 5 in the test function {5≤⊃⌽⊃⍺} is of course the ceiling of the square root of 20, the length of the bit mask. There is no need to calculate it on each iteration. But it turns out I can calculate it once and the test fn can read it.

      ]disp { L←⌊(≢⍵)*¯1 ⋄ {⍺,1+⍵/⍳⍴⍵}/(⊃{p←1+⍵⍳1 ⋄ (⍺,p)(⍵∧(≢⍵)⍴((p-1)⍴1),0)}/)⍣{L<⊃⌽⊃⍺} ⊢ 1 (0,1↓⍵⍴1) }20
┌──────────────────────────┐
│1 2 3 5 7 9 11 13 15 17 19│
└~────────────────────────→┘

Cleaning that all up...

      sieve←{⍵∧.=0:⍺ ⍵ ⋄ p←1+⍵⍳1 ⋄ (⍺,p)(⍵∧(≢⍵)⍴((p-1)⍴1),0)}
      es←{ L←⌊(≢⍵)*¯1 ⋄ ⊃{⍺,1+⍵/⍳⍴⍵}/(⊃sieve/)⍣{L<⊃⌽⊃⍺} ⊢ 1 (0,1↓⍵⍴1) }
      es 20
1 2 3 5 7 9 11 13 15 17 19

What are the scoping rules that make L visible here to the test function?

10:38 AM

@SJT Lexical. L is defined inside es so everything es defines can see L.

@SJT Btw, no need for that ⊢

So dfns have the same scope rules as tradfns? (I thought I remembered there is a difference.)

Tradfns use dynamic scope. Everything I call can see what I've defined.

@SJT (≢⍵)*¯1 is ÷≢⍵ and ⍵/⍳⍴⍵ is ⍸⍵

brgal

@Adám Thanks Adám !

How is ÷≢⍵ square root?

How is (≢⍵)*¯1 square root?

Did you mean (≢⍵)*0.5?

@SJT 0,1↓⍵⍴1 is ~⍵↑1

10:44 AM

Ouch, yes.

Nor do primes start at 1.

      { L←⌊(≢⍵)*.5 ⋄ ⊃{⍺,1+⍸⍵}/(⊃sieve/)⍣{L<⊃⌽⊃⍺} 2 (0,2|⍳⍵-1) }20
2 3 5 7 11 13 17 19

Tx. Good to catch up a bit with Dyalog!

2 hours later…

pVC aecidiospore adduced

1:03 PM

@Adám so in 18.2 we migt be able to use dyalogscript?

1:15 PM

@pVCaecidiosporeadduced Yes.

1 hour later…

user

2:21 PM

@Adám Why is that? Were there problems with 18.0 or is 18.2 supposed to be separate from that, kept for backwards compatibility?

rak1507

2:36 PM

@user dyalog.com/blog/2021/06/…

user

Thanks!

3 hours later…

5:40 PM

Announcement: APL Campfire in 20 minutes on Zoom. Passcode: The result of ×/1966 11 27 evaluated as APL.

3 hours later…

APL Campfire ft Geoff Streeter and Andy Shiers

8:48 PM

Announcement: The APL Campfire event featuring Geoff Streeter and Andy Shiers is now available on YouTube:

►

2 hours later…

10:19 PM

@Adám You remember helping with with the ⎕LX expression a week or so ago?

I managed to get it working, but now I can't open the dang workspace! :C

]load 'my.dws'
* Command Execution Failed: Invalid UTF-8 sequence

because the ⎕LX is expecting a string, but I don't know how to open the WS without firing the ⎕LX expression

10:53 PM

@nathanrogers What did you search for on APLcart?

⎕LX, I found the command line option -x in the docs

but I flat can't even open the file

I'm not running "dyalog my.dws -x"

I have dyalog open and trying to ]load it

aplcart.info?q=open%20without%20%E2%8E%95LX

I also looked for ]laod

]load is for SALT files. Workspaces are not SALT files.

ws not found

)xload my.dws
ws not found

)xload 'c:\absolute\path\to\my.dws'

same thing

10:56 PM

Use double quotes. (Or no quotes, if there are no spaces etc.)

oh, well why is that the case?

Because nobody asked me how it should be designed.

:P ok well thanks for that

I did look, but I couldn't get it to work

I should add an example

I can't say I've ever had to use double quotes

11:01 PM

Yeah, it is a mess. I my opinion, all commands should just go away, and be replaced by functions where appropriate.

18.2 will have a new ]get command (!) which should work for most cases of wanting to get anything from anywhere.

rak1507

@Adám suggestions for the cart, unix timestamp related things could be handy