@AlexanderBelopolsky nice! very similar to what i had in mind :)

@AlexanderBelopolsky can you find a better way to do ,/{(. x)#'!x}? ;)

I wish we had an x!y to (x;y) primitive (I think we will)

use ¡ :p

well, that clashes with "scan". one way to avoid this is to reuse the syntax for "repeat" (is that what n f/x is called?) with a negative count: -1(!/)/x in order to express the inverse of !/

but implementing this would probably be too complicated...

apl and j do it (⍣¯1 in apl and ^:_1 in j) but they are more tolerant to bloat

@AlexanderBelopolsky (#:'=:)' -> #:''=:' is a byte shorter

I argue that fold and scan are not useful for non-associative dyads and can be given alternative meaning

@ngn - I still like count each group better.

@ngn - -1(!/)/x is way too clever! Would it make sense for anything other than ! ?

@AlexanderBelopolsky -1(`sha@)/x :)

@ngn - Nah! That would be `sha?

ah, right... :)

@ngn - groups.google.com/d/msg/shaktidb/UoxISHE8GMw/sdb8S460CwAJ

@AlexanderBelopolsky & - exactly! :)

 f:{&0|-/(#:'=:)'(x;y)}
 f["seventeen";"there"]
"seevnn"

btw, i still don't understand the need for : in #:'x

my impl doesn't require it

any faq suggestions? k7contrib.gitlab.io/docs/#faq

@chrispsn summary of date handling, maybe some crypto examples would be interesting, differences from q (a complex subject)

@ThomasLackner thanks!

will probs also include these once we have answers groups.google.com/d/msg/shaktidb/UKZcbwN-qiA/i3LtReSPCwAJ

@chrispsn file and network i/o would be cool too :)

@ThomasLackner that and bringing in typed cols from CSVs

what are the columns in trace?

 \u:10000 sum !10000
|n u t
|- - -
152

Any strong views on making this work better on mobile are appreciated but feels like it’ll be a stretch if we continue down the “clickable index” route k7contrib.gitlab.io/docs

q’s docs look great on mobile code.kx.com/q4m3/12_Workspace_Organization

@ThomasLackner done. any idea how to use the two .z functions on the IPC front?

@chrispsn this still seems to work for me (but only with port, not host:port) groups.google.com/forum/…

@chrispsn first of all wanna say im very impressed by your effort on this so far. to your q: long term i think one-page-per-topic with lots of interlinking is the only way to go. should probably generate them with k too. that help splash on the left is kinda puzzling for noobs anyway - maybe best to deemphasize. my two cents

Also,

#[1 2;]
,(::;::)
2+
+[2]
+[2;]
'typ
+[2;]
^

@ngn - what happens here:

1 2#'
(#[1]
 #[2])