@Adám With behind, any better way to write your {u≡⍋⍋u←∪⍵} then as (⍋∘⍋⍛≡⍤∪⊢) ?

Or is that the best you can do given the (behind unique) gives you an atop

 LENGTH ERROR
       a b/⍨←⍬
          ∧

I'm not understanding something :|

Wait... Somehow I got confused, thinking that ⍬/1 2 3 worked.

what does mapl stand for in the Dyalog installation?

I have many rows of character vectors that look like this: "12/02/2022 11 23 54 22 6" - I've been trying to figure out how to turn these into many rows of 6-element vectors instead. I've tried the APLCart, searched the manuals, etc.. This should be easy right?

@cube.head Does this give what you want? (⊢⊂⍨1,' '∘=)'12/02/2022 11 23 54 22 6'

If you don't like the trailing spaces, one way to strip them is like this: (' '~⍨¨⊢⊂⍨1,' '∘=).

@B.Wilson or just (' '∘≠⊆⊢)

@RubenVerg Uhhhh. That's way better! :D

@cube.head If you're doing actual calculations on the data or whatever, then I'd argue that turning this into an 8-vector of numbers is perhaps more convenient: (⍎¨~⍤∊∘' /'⊆⊢).

Thanks.. I'm not sure I'd ever get this

does that sound accurate?

yup

Yeah, though that's mostly just an explanation of how forks work. The key is really in how Partition (i.e. ⊆) works.

hoo boy. Please tell me that we all start out this confused :-)

thank you again guys

@cube.head You're in very good company :D

If you want a more details breakdown of the previous code, or if you just want pointers to specific reading materials or whatever, let us know!

@B.Wilson thanks.. I will study and experiment on just this expression for a while

" have many rows". For any sizeable value of many, partitioning each is not going to work well.

Let alone an execute each inside a partition each. Look at ⎕VFI.

       a←1e6⍴⊂'12/02/2022 11 23 54 22 6'
       cmpx 'z←('' ''∘≠⊆⊢)¨a'
 3.9E¯1

@PaulMansour Will do, thank you

@cube.head The above shows ('' ''∘≠⊆⊢) running in half a second on a million records. That's pretty reasonable for government work.

@B.Wilson Yes, I have no more than 15k rows - I guess I should have defined "many" :-) ... VFI is interesting though.

Yeah, just starting out, I wouldn't overload too much on a billion different concepts. My pedagogical advice, FWIW, is to stick with learning the primitives for now.

If you have clean data, you don't even need ⎕VFI, just replace the slashes with blanks, ravel, execute, and reshape.

No each, no nesting.

@PaulMansour I like the way you think! Do you have much experience ingesting "large quantities" of data?

Somewhat related, I've been wondering how to ingest stream data in Dyalog, given a pipe or socket or whatever.

@PaulMansour Those things at column 1 of each row are dates

@cube.head That's fine if you always know that the first 3 numbers specify a date.

Is there a functional difference between Disclose and ⍬⍴?

@cube.head Yes, I assumed the whole text vector is a time-string with hours, minutes, etc.

@B.Wilson "large" being up to about 100 million rows or items. Not in the billions.

@Silas That'd be it.

@B.Wilson Yes, he's lead dev of a db product for "massive amounts of data analysis" implemented in Dyalog APL.

@B.Wilson You should be able to tie it and read chunks.

@B.Wilson Yes, ⊃ discloses the first element. ⍬⍴ returns it enclosed.