« first day (1600 days earlier)      last day (1044 days later) » 

10:06 AM
CMQ: Which APL expression is it typing?
 
 
4 hours later…
2:13 PM
@Adám wow TIL: selectric typewriters
reminds me of the stargate routing to a planet
 
2:40 PM
Hello
For weeks now, I have been thinking about ways to render N-dimensional arrays in a nice way (where N>2).
I have a prototype version of KAP that uses graphical rendering of arrays instead of using block drawing characters. This should allow me to be much more flexible when it comes to rendering higher-dimensional arrays, but I just can't think of a nice way to do it.
If anyone has any idea, I'd be happy to hear them. I've even considered using OpenGL to render 3D arrays in, well, 3D, allowing you to rotate it.
(that's a funny idea, but not practical, so I may yet implement it as a joke)
 
@EliasMårtenson i'm fairly sure many people have. I somewhat like what dzaima/BQN does (still with box drawing chars of course), but not sure anyone other than me does
 
@dzaima the BQN one is quite simple, and doesn't seem to do much at all for higher than 2-dimenional arrays?
I mean, it separates them with some number of newlines, but that seems to be it.
 
@EliasMårtenson that's marshall's formatting, which is quite different from mine in dzaima/BQN
 
Oh, how can I see yours?
 
(mine is comparatively extremely complex and weird)
 
2:48 PM
it might just be simpler to have a boring representation of arrays generally and have a view function for opengl
 
I don't have dzaima bqn installed. I mostly play with it using the web application.
 
@EliasMårtenson there's on online app?
where?
 
some things
https://dzaima.github.io/paste/#07VdLTsMwEN3nFFZvYNYcCG6RFimLgkRFgkCCXekKdukuSL1LTlKnTlyPP/FMGKSCIo1ae/r6Zuz52BZCXLW7tcxkmx9kJrppftCqtnpsq@UgpZrfSSHVJxi9WJi3M4FNUjxZGCU1nA7kjr2gRuN66844/RvwNahxFqSl8TTOQsmLLcoLWezW9e753vP38@8HZyxAMalxMDt7EbEnIU@8uFwhITM/Bs54/Fe2/6djTkJqXlROk5DGX1wc0Mj/HQdUrZKQwQYQkwaN7BvGbn29uLldKCMPp/rMB/myxko2CrHqkOcvpakgaA@nr/bRijZTlJ6Zdwj6htOPn7kG3cO0yhrfHhOrRWL6lgjXGNFpu4kdQ2I018q3G9BpbGLXkRhkyjf4NIdn4i8EuzuG5mBPDPY2vceBQ2oZu7TNcbi8oqMXHl5qPqqxi@3E3sDOZi7IPHnOzpYZUlTtkNHmAsdTb@xsxj9cDZPRhp@n7tnZIg@SqT2dnc3sH1/9sLLN9TPXD@NZzM5GepDyPlpxD9sj#BQN
 
Thanks.
Nice and consistent.
 
(well, way less consistent than marshall's)
(like, compare these - those are both rank 3 arrays with the same shape, but are displayed completely differently)
(my goal wasn't to be consistent or simple, but rather to represent the data compactly and possibly in a more usable/readable form, while keeping it unambiguous)
(idea being that the thick arrows are like the thin version, but with a merge afterwards (BQN >/Dyalog /GNU ))
 
3:10 PM
@Razetime there's no online app for dzaima/BQN, only marshall's JS impl
 
3:23 PM
@dzaima Yeah, you're right. There isn't really any need for the inner borders in the second case, is there?
 
@EliasMårtenson I guess not strictly. I still like some separation between non-primitive objects though
in that model though, ones in shape become unreadable
 
with the virtual keyboards for dyalog apl and bqn I was wondering: are there reasons they are laid out like they are? (I assume so)

e.g. the order and grouping of symbols

I see that bqn has some groups colored (that helps also I think)
 
@dzaima i guess it could still look like this, but i don't know if that's worth it
 
being fairly new to the family of languages I was wondering if it would make sense to put the symbols next to the repl instead of above it and also whether it would make sense to group them somehow, perhaps even in some "fixed" way

in the dyalog ui and also on some of the web based APL like uis I found often the symbols break into a new row depending on how wide the window is
 
@tosh Like this?
 
3:36 PM
@Adám this is great!
I guess once you are more familiar with the symbols you tend to hide them or don't need them grouped
 
Right.
Also, it is really hard to definitively group the symbols, esp. because some have no connection between the monadic and dyadic form. E.g. is Match and Depth.
 
@tosh BQN's starts with functions (basic arith, some logical arith, comparison, structural fns, indexing), then modifiers (with related 1- and 2-modifiers besides each other (i.e. ⎉˘, ⚇¨, ⍟⁼, (which all also have the same key))), then general syntax
 
as a beginner I really like them present at all times as reminder but also because they make me get new ideas or help me explore so for learning having these bars are great not only for input (just a random observation)
 
@tosh yeah, I've found myself going through the language bar while thinking about ways to golf
 
@dzaima ty for the clarification
 
3:40 PM
@tosh Have you installed Dyalog locally?
 
@Adám yes, re-installed it yesterday
for some reason I prefer the tryapl ui (I'm on a Mac) but can't put my finger on why yet
 
(Dyalog groups them (assignment), arithmetic, number theory (+ left/right), comparison, selection, lookups, replicate/expand/reduce/scan, structural, operators, task-specific stuff, syntax)
@tosh If you do figure it out, please let us know so we can improve RIDE. Also, if you want the language bar grouping to be different, you can simply drag the symbols around.
 
ah interesting @ dragging, thx for the hint
 
 
1 hour later…
4:53 PM
@tosh janiczek.github.io/tryapl-elm has groupings
and also the hovering thing
 
@Razetime I meant the web version of BQN that is available on the main site.
 
which I found really useful locally when I was starting
 
@rak1507 But same groupings as Dyalog's 3 interfaces. Question is why those groupings. I don't particularly like them. See my proposed alternative.
 
True
That alternative is much better imo, has there been any feedback on it or plans to do it?
I see it's from 2017...
 
@rak1507 No, as you can see (it would have been on that thread). Change comes slowly…
 
4:57 PM
everyone seems in agreement yet it doesn't happen for ~4 years? :(
 
We need more devs.
Some people probably want the traditional layout, but if someone was to volunteer to add an option to RIDE that lets you choose an alternative layout, it'd probably be accepted. Notice how there are no outstanding human-PRs.
 

« first day (1600 days earlier)      last day (1044 days later) »