APL is after all the ultimate language. Writing complex programs in just one line of code is rather remarkable.

All I can say is that you're not alone

Thanks :)

oh man, that was my project in the first year of computer science, can't say it was easy. the voices still haunt me to this day

‎

You're hooked. You can try Q, K, even J. But you'll get back to the symbolic high. ;-)

Actually, this one’s not that hard to understand because the algorithm is straightforward. The APL code for sudoku is much harder, because the algorithm there is non-trivial.

When I learned Haskell I used: www.edx.org/course/introduction-to-functional-programming & book.realworldhaskell.org/

@@code_report Thanks! And what text editor/ide do you use to write code in Haskell?

@@Luj8n I usually just type up solutions in GHCI if it is simple. Otherwise I use VS Code.

For APL I use RIDE: github.com/Dyalog/ride For BQN I use: bqnpad.mechanize.systems/ or mlochbaum.github.io/BQN/try.html I am learning APL, J and BQN all at the same time. Probably BQN is my favorite. I just don't know it as well as APL at the moment.

Emacs has a pretty nice APL mode too, including REPL support (at least for GNU APL, never tried with Dyalog). It also has an input mode for the APL glyphs (C-\ APL-Z ret) by prefixing them with . and then pressing the correct key for your symbol afterwards. It comes with an overview of which glyph is on which key after the prefix too.

At the moment, I find APL more beautiful. However, you are correct that J is a superset of the ideas/genius in APL. I am learning both.

@@code_report I'd love to see your J version of this same video

I learned J but haven't learned APL. I found this video to be intelligible from my J experience though, so there's a decent amount of crossover. It might be worth learning J first purely on the grounds that it's more approachable and then use that understanding to go back and learn APL if you want.

APL is based on Ken Iverson's mathematical symbols, not a typewriter. He used Greek letters, as well as special overstriked characters from an IBM typeball made especially for APL.

You need to set quad IO to zero. Run: ⎕IO ← 0

@@code_report thanks, awesome. Can I also ask why this needs to be changed, what is actually occurring (philosophically speaking I guess)?

@@oraqlle By default, APL uses arrays that start at index 1. This can be changed however by switching the index origin (⎕IO) to 0. This index origin affects the roll operator as well so setting IO to 0 will change roll from starting at 1 to 0.

@@loganalt1838 Thanks for the explanation. This got me scratching my head also.

These rules are equivalent to the original rules. The original rules as taken from wikipedia are: - Any live cell with two or three live neighbours survives. - Any dead cell with three live neighbours becomes a live cell. - All other live cells die in the next generation. Similarly, all other dead cells stay dead. Condition for a cell to be alive: - Already alive with 2 or 3 live neighbours - Dead with three live neighbours Considering itself as a neighbour the condition becomes: - Already alive with 3 or 4 live neighbours (+1) - Dead with three live neighbours Or in other words: 3 live neighbours OR (self and 4 live neighbours) This is equivalent to the APL expression 1 ⍵∨.∧3 4=... Hopefully this makes it more clear

@@rorykemp5218 Thanks for the detailed explanation! I didn't try to check the logical equivalence as it looked too different... now it makes sense 😊

@@japedr no problem, glad it makes sense now

When the 9 rotated matrices are sum reduced, this includes the previous board state. Because of this, when he checks for 3 neighbours, this then includes the case of a previously live cell with 2 neighbours.

By default, APL sets indexing to start at 1, but you can change this by using ⎕IO ← 0

I believe Dyalog's default indexing starts at one. To get the behavior in this video you need to run ⎕IO←0

Thank you very much for the information!

I believe they are not copied in memory eight times. It's a functional language: lazy computing, memoization, and all that.

Fair enough. Is there any way to verify that?

​@@korvmakak7727 APL is not lazily evaluated or memoised, so it does indeed create 8 copies. Using wsreq dfns.dyalog.com/n_wsreq.htm it can be seen that applying the rotations on a matrix with a million elements (1000 by 1000) takes just over 1 MB, which is 8 times the size of the matrix considering APL uses 1 bit per boolean. Interestingly, summing the matrices to create the neighbours count then uses 3 times that, it must not be done in place for some reason.

@@rorykemp5218 Wow. Thanks for the detailed answer!

Also, I learned APL as my first language, back in 1973, and my work in cellular automata is very well known. See the papers titled Computational Ontogeny

This implementation seems to behave exactly right, with gliders and everything. I don't know if that would be the case if he used a different ruleset..

@@jessevos3986 See Golly. Create a pattern with a home cell in the middle, and four neighbors, one at each corner of a neighborhood that is 3 by 3 cells in size. I will put my faith into Golly, which is the premier cellular automata software, rather than your assertions. Just so you know, I learned APL at Golden West College and Orange Coast College, from such luminaries as John Clark (check his history). If these admonitions are not good enough, then please refer the the original publication of John Horton Conway's model, as published in (as I recall) the October 1970 issue of Scientific American magazine.

@@vNCAwizard You did not understand what I said. Also, please think about something yourself first, instead of blindly believing what anyone says (they could have made a mistake), or what you think is right (it is possible to misremember), or what some program says (you could have configured, or interpreted it incorrectly). Consider what evidence reality presents you (the implementation in the video behaving exactly how it should) and compare it to your beliefs (the implementation in the video is incorrect), after that, you can start figuring out exactly where the disconnect is. Unless I am completely misunderstanding what you're trying to say, which is possible, in which case I excuse myself for coming across rather demeaning.

@@jessevos3986 Try Golly before you assert that four on cells result in creation. I have you nailed. Try Golly. Then review your assertions in prior videos. I don't forget very much.

Mo bamba