Oh boy, I'm dreadfully confused about this dataype... the short summary is that ⎕DR = 326 Pointer (32-bit or 64-bit as appropriate). Who knew there were pointer types in APL?!?!
@AviF.S. what is (1 2)(3 4 5) if not an array of 2 pointers to 2 other arrays?
my non-⌺, probably reasonably inefficient, solutions in dzaima/APL because why not: https://dzaima.github.io/paste/#07ZOxilRBEEXz@YoT7qJgV1d1db9k843M/BkRZBEV9S2zMKKBocEGZruDIEbul2x/iV3OBKs4sMGABgOPx6XrQtWBe09Q9Gk/f3/6ePyenPaXy0Siv33B0VGfr/r87fjB7XJ9e/GaoW8@bES833zt82Wfr/u8Pqa/eTVcY7J@NiH51/NVk0Vfveur5zu@L/seLce@MwQyaEiDAh6yQoNpSEmIxJGrs/CLIoaUkI5UpIWcyIksQ@ZMVrJt/bmQnVxDNvKEpiFV0IzGVjW0oL7xf9p97vd9jz7HfVrRhgapJUywIDXFDAtKc6xibctjEyVRgrRkilKCtBSKU4KyNMqEp63fBc94kLrhBQ9Sr3jDY2tNVKHmf8RflWrUIK1OrdQgrRMt0eT3gCzuE5I7AVncJyR3AjL8q93nXu979HFx8rc@y599Hh1@@OPy0abRhzYf2nxo8//Y5p8#dAPL
@dzaima Huh, I wouldn't have thought about it that way. Surprise that APL does. Well, it's my fault for not having used ⎕DR before. There's still more that I was about to post, as it doesn't completely answer my question. But thanks! It certainly helps!
Longer version: It's generated by my attempt to replace letters in a string with their indices in the alphabet. Of course, this isn't correct because the numbers have to be ⍕, but curious what it's doing
@dzaima I already have a stencil solution, in a single line it becomes {(⍺>1)↓({⊂⍵}⌺(↑2/⊂⍺)) ⍵↑⍨ -(⍴⍵)+ 2× ⌈2÷⍨¯1+⍺} so you can code a ⍴↑↓⍉ solution and then compare the two :P
Other clue is that I can append characters to it and it prints the same way, as if it really is a character vector, but the length, display & type aren't matching those other clues...
The other way it differs from a character vector is that you can't apply ⍎ to it directly, you have to do ⍎⍕ to get the array represented...
@dzaima Ugh, it's so painfully obvious now I can't bear to look at myself...
@dzaima I was so terribly confused...
What's a better way to figure out the type of something? There must be a way to have seen that it was a mixed vector of characters and number, and even which was which, no?
@RGS Not only did that not work, unfortunately, but there also might be something more ideal than seeing a list of numbers representing data types, no?
@AviF.S. seems the best dfns option is still to look at ,¨ - notice the ~s at the bottom. Sadly it doesn't do anything cool for mixed-type simple arrays
There is an alternate set of rules where I can roll 2D6 once every year, to get one skill. What should be the one-year roll value to have the same expectation of skills as rolling 2D6 for 8+ every four years?
I think so, but I believe you've got a sign error and a fencepost error somewhere; AnyDice says that the probability of 8+ on 2D6 for a single roll is 41.67%
... which is consistent with AnyDice, so I'll take the code as correct. The "hold on" was more so I could try to work through the code and understand what was going on.
in the new ruleset I get to roll 2D6 each year regardless of what happened in previous years and if I roll "well", I get one extra skill regardless of how many skills I already earned
ok. so by the time 4 years have passed, the number of skills you earned is obviously the number of skills you earned in the 1st year + skills earned in 2nd year + skills earned in 3rd year + skills earned in 4th year
but the nr of skills you gain each year is something that is completely unrelated to what happened in the prev years and what will happen in the upcoming years
That implies, then, that if I have 100% probability in four years of one skill (I don't have to roll, it's automatic), I should need a 9+ rolling every year.
@JeffZeitlin I see, but if you took the 9+ lower, to things like 7+ or 8+ then you'd be making it really easy to earn a skill in a single year, so that by the time 4 years have elapsed, you earned a lot of skills :-P
@Adám If it had occurred to me it were a mixed vector, this would've been a beautiful display of which were which! But is there a generalization I'm not seeing here, for what you're suggesting in the future?
Well, in writing my own rule, I can theoretically do anything I want, but there are certain traditions which will meet significant resistance if broken...
Yes. Most of the time 2D6, occasionally 3D6 or 1D6
Later versions and mass combat rules might allow 4D6 or 5D6, but it was more likely that they'd change the damage level rules and muffin it back to 2D6
hmm if you were allowed to propose crazy rolls, you could do "roll an odd number and then roll something larger" for 25% (equivalently, roll an even number and then something smaller)
@AviF.S. Uh, yes, this is a general solution. It will take any array and recursively replace all numbers with 1 and an all characters (and other simple scalars) with 0.
@Wezl I don't mind answering, but I'm not always around. :-) That said, please let me know if you try searching for something and don't find it, whether with your initial search terms or at all.
@RGS When entering a function that can see its environment, but also can shadow some of the surrounding symbol table, there must be a new symbol table. The locals' declaration defines that symbol table.
@RGS Hm, I don't think there is any short way to do it, other than get the ⎕NR of the function and remove all comments and strings. Remove up until before the first ; on the first line. Locals are then all names following ;.
I meant the syntax you use to code it, I never wrote a tradfn in my life and the stubs in the contest file don't have local variables enumerated so I can't really work from that
@Adám I don't think so... I searched for "tradfn" and a bunch of results popped up and then I didn't read the descriptions, I looked at the code samples to look for a tradfn header, that I thought started with ∇
When I saw no code sample had a ∇ I assumed I was not going to find what I wanted.
"a bunch of stuff to explore in ⎕SE" means what, exactly?
i.e. how do you explore a quad thingy?
What I mean is: by "a bunch of stuff to explore in ⎕SE" do you mean the source code of ⎕SE? I don't know if it makes sense to talk about the source code of ⎕SE
@RGS ⎕SE is a root namespace, just like #. You can navigate to it with )cs ⎕SE or just type ⎕SE. and use AutoComplete to explore. Set Options>Configure>Auto Complete>Make suggestions after: 0. Or you can open the Workspace Explorer (second button in the "Tool" toolbar). Or use ]map ⎕SE (huge output!).
Apart from that, I looked into ⎕SE.Dyalog.Utils and I noticed that in the display function something named ch is used but it isn't defined in that file (i searched for it with the search bar) so those files have interdependencies?
@dzaima Not really. The huge enormous fundamental design mistake was using the same symbol for both opening and closing the editor. If they had only used two different symbols (e.g. ∇ and ∆) the world we live in would have looked very different.
@Adám imo planning for an inline editor in the REPL itself is a problem. But either way, the only distinction between a tradfns header and source being that the header is only at the first line is bad enough.
@Adám Every other practical language I know I'm pretty sure has a way to tell syntactically whether a line defines a function or not. In APL a←b c d can be either a header or assignment.
