Hey @all. I have been trying to figure out how to erase my session history. Right now I have 3k+ lines of stuff in RIDE I'd like to erase but haven't found anything on how to do that. Any clues?

Asked too soon! Just found it in /home/.dyalog

@WilliamA.Noble Welcome to the APL Orchard. Happy you found the solution.

@Adám Thank you, Adám. I'm really excited to learn APL, it's really like nothing I've ever seen! I'll come around here from time to time

@WilliamA.Noble Great. Can I ask which learning materials you use?

As of right now I've read the basics in the "Learn" section of tryapl.org. As soon as I get finished with that I'll try reading the xpqz.github link in the sidebar. From what I've seen it seems very good.

I also consult help.dyalog every so often

I'll accept recommendations, though

xpqz's book is good

@WilliamA.Noble Those are all good, but you can also check out apl.wiki/Learning_resources

Thank you for the resources, I'll check them out

@WilliamA.Noble I'm always interested in any feedback you might have. Thanks for checking it out.

@WilliamA.Noble I have a desktop shortcut which does sh -c "rm -f /home/palaiologos/.dyalog/*.dlf"

could someone explain 10 20 30 ({⍺ ⍵}⍤¯1)3 4⍴⍳12 to me

Sure.

⍤¯1 means {⍺ ⍵} is applied on each major cell of left and right argument together, so we get an array where the major cells are 10{⍺ ⍵}1 2 3 4 and 20{⍺ ⍵}5 6 7 8 and 30{⍺ ⍵}9 10 11 12.

Since each function application results in a 2-element vector, we get a 3 by 2 matrix result.

@KamilaSzewczyk Makes sense?

hmm, yeah, i see. i was really confused by rank -1

Neat, also yeah what's up with ⍤¯1?

It simply makes the applied rank relative to the rank of the respective argument.

Left arg has rank 1 and right rank has rank 2 so in this case ⍤¯1 is equivalent to ⍤0 1

Awesome.

I.e. we pair up scalars (elements) with vectors (rows).

the concept of ⍤ is closely related to monadic ,

is it?

How so?

⍤ is much more complex, monadic , simply vectorizes an array. I guess you can use ravel + bracket axis in similar ways as ⍤

I mean ravel order

Ah

Ah, yes, very much.

See also the interactive diagram at apl.wiki/Cell#Characterization

wow, that's cool

one day someone asked me about what's the difference between NumPy's broadcast and APL's rank operator

What was the answer?

I deduced that broadcast is implicitly reshape the smaller array to the same as the larger one

Yeah, that sounds right.

while rank is like the for loop in C (according to my understanding to J)

But they have to conform, right? I.e. the smaller array's shape has to be a suffix of the larger array's shape.

What clicked for me was to think of rank as a generalised 'each'

Yes, the shape has restriction

@xpqz That'd actually be the depth operator. I've been tasked with pitching, speccing, and modelling that.

18.2?

No, 19.0 at best.

@xpqz yeah, or a generalised map/iteration construct

The rank operator is more like generalised scalar extension.

I'd really like for Dyalog APL to use leading axis agreement too.

I like to think of outer product in terms of rank, even if it's a bit slower.

yeah if you have rank you don't need outer product

Yeah, well outer product is just a special case of rank (and each).

And each is under disclose rank.

sometimes rank is more efficient than outer product too as it can avoid constructing the full matrix

Hey @Adám, I just handed in a thing for the cataloguing trial run in case you are still interested. Sorry for doing this ridiculously late, if you must know my parents are renovating so it is hard for me to focus on stuff.