the k tree

« first day (77 days earlier) ← previous day next day → last day (1458 days later) »

juanez

00:42

@chrispsn I created them, same as in q

10 hours later…

chrispsn

10:18

from latest \h update: asof <0 3!10 20

hmm

chrispsn

10:46

20190522
?2 3 5
^2 3 5

that could explain why sort is now a primitive (deferred)

Possibly dumb question: is it impossible to define new infix functions in k? So eg 1 {x+y} 2

ngn

@chrispsn afaik it's impossible. lambdas are nouns

chrispsn

@ngn thanks. and tacit doesn't get you there either? "hello" (^:,) "world"

ngn

@chrispsn parens nominalize (turn into a noun) whatever's in them

chrispsn

ahhh

Adám

@ngn @chrispsn But is there anything syntactically preventing this? IIRC, there is a hacky way to do so in K4 by adding your function to .q which is how Q's infix functions are made.

ngn

10:57

so it parses like "hello" ((^:,) "world"). the part on the right evaluates to a projection and then "hello" is indexed with that projection, so the result consists of char nulls, i.e. spaces

why 2 spaces instead of 1 ... i've no idea

Adám

Ah, Q does allow defining infix functions.

Is Q on topic here?

chrispsn

Sure

@ngn just to complete what you said:

 `p "\"hello\" (^:,) \"world\""
"hello"
((.;^:;,);"world")

ngn

@Adám the problem is when you name a dyad f:{y} and use it later 1 f 2 - the parser can't know that f is a dyad (without doing some sort of analysis of the whole program or using runtime information like in apl), so it's better just to assume that all names and all lambdas are nouns

@chrispsn interesting. . is used to create a composition. it was ' before

and that k7 output formatter really sucks...

@Adám what i used to point out to you as a big advantage of k5/k6 is now gone in k7 - the ability to copy the output, feed it back into the interpreter and get the exact same result :(

chrispsn

the REPL occupies this weird spot between being an exploratory tool and... the main way to access the language

i really like things like tables being represented as tables

but coming from a spreadsheet background, it feels strange to not have my past work persist

@ngn wasn't it suggested that there may be a flag that starts the interpreter in a "raw" mode?

or there could be in the future

ngn

11:27

@chrispsn i don't know

@chrispsn with proper indentation, an expression generating a table could still be made to look like a table

 +`a`b!(1 2 3;4 5 6) /currently printed as:
a b
- -
1 4
2 5
3 6
 +`a`b!(1 2 3;4 5 6) /it could have been something like:
+!/(`a`b
   +(1 4
     2 5
     3 6))

chrispsn

@ngn at least there's `k

ngn

11:44

@chrispsn i see the repl as something ephemeral. work worth persisting should be done in a .k file

@ngn oh, i could have used ![..] instead of !/(..)

chrispsn

12:02

is it normal to always get the same rand results on startup?

(base) chris@chris-VirtualBox:~/Desktop/test$ k
2019-05-23 15:34:17 2core 1gb avx2 © shakti l2.0 test
 rand 4
0.5 0.9078156 0.269656 0.5607996


(base) chris@chris-VirtualBox:~/Desktop/test$ k
2019-05-23 15:34:17 2core 1gb avx2 © shakti l2.0 test
 rand 4
0.5 0.9078156 0.269656 0.5607996

Adám

@ngn I hope we can get that in Dyalog 18.0 using the new array notation.

ngn

@chrispsn yes, the initial prng seed is fixed. \S is supposed to change it but it looks like it's not done in k7 yet

3 hours later…