« first day (738 days earlier)      last day (797 days later) » 

00:04
re: developing a K "mindset", a very good resource is the archives of old kx.com kdb and k listboxes and forums. Them may be available on archive.org. For instance web.archive.org/web/20061101073453/http://www.kxforums.com/kxf/… but I think list boxes were more interesting.
 
2 hours later…
01:58
@bakul thanks for the link
 
2 hours later…
04:09
The listboxes: https://web.archive.org/web/20060704054229/http://www.kx.com/listbox/k/index.html

https://web.archive.org/web/20060715025722/http://www.kx.com/listbox/kdb/index.html
2
 
4 hours later…
07:51
@ngn i don't accept. i will write one, one day.. being able to place a dictionary somewhere and get a property dialog, or a table (for a table), a coulple of buttons calling k functions. that should be doable. than i want to be able to write simple gui programs in plain k with minimal overhead.
 
3 hours later…
10:35
@ktye mesh does some of this
Drag and drop data structure is v much doable
 
5 hours later…
 
5 hours later…
ngn
ngn
20:36
@ktye that would be nice but it would be useless if the underlying k is not free
@ngn i'd probably write it for one of my future k's in go. but i'm not in a hurry.
ngn
ngn
@Traws "O(n) sorts" - a white lie
20:52
@ngn sorting any finite set can be O(n) (and if not assuming bounded integers, int comparison sorts would be O(nlogn*log(max)) anyway)
ngn
ngn
21:05
@dzaima sorting any finite set is O(1) :)
@ngn an unknown length list of elements from a finite set is what i meant. There is a minimum O(n) requirement just to go through the list of elements
ngn
ngn
the definition of O includes lim n->∞, it doesn't work for finite inputs. but that's just strictly speaking.
@ngn the "unknown length" part can go to infinity. But you'll still be sorting 32-bit ints even if you have a googol elements
ngn
ngn
@dzaima ok, right
counting sort is the obvious way to get an O(n) integer sort, but you can do much better, e.g. radix sort, which is for all intents and purposes O(n) but doesn't use 16GB RAM
22:06
implemented a basic base 256 radix sort in java (sorting 0 to 2^31 because handling signs is harder), results
ngn
ngn
23:02
@dzaima in practice it could pay off to make an extra pass over the input to determine min and max. if they are not too far apart, radix sort could require fewer iterations. that's what i did in ngn/k.
@ngn right. (been meaning to play around with sorts for a while, this conversation finally pushed me to start)
ngn
ngn
@dzaima i wish you were doing it in c, to share code :)
@ngn I intend to just find things that work well in the first place, proper performant impls come later (who knows how later), when i have an actual impl to put them in

« first day (738 days earlier)      last day (797 days later) »