« first day (2945 days earlier)      last day (3 days later) » 

05:54
Very excited to hear from the Array Cast that the APL386 font is getting redrawn! I've always been a bit bothered by the irregularities in the old version.
 
3 hours later…
08:38
@rabbitgrowth You can try the latest version (ttf only) at abrudz.github.io/APL386 but be aware that the currently most obvious issue is too large line height.
09:13
Nice. How close is it to being finished?
Very close. It is the only remaining issue (I know of), and it is a regression from the previous version.
Cool, looking forward to the final version!
09:38
@PaulMansour I think it's fine to see bracket axis as "syntactic sugar" over using the rank operator with dyadic transpose, but using rank and transpose is a good way of understanding what enclose-with-axies ⊂[] and merge axes ,[] are really like
10:24
@Adám hmm, not sure what I think of the new epsilons
@RubenVerg Per Aaron Hsu's request, to make more distinct from e and ϵ.
is there bqn support?
@Adám hmm
I think I might get used to it
diaereses look closer together, that's nice
thin tilde is weird but I think I like it
no TinyAPL support I'm guessing, might make a PR(:
@RubenVerg To the best of the designer's ability. The font contains double-struck letters (just the ones needed for BQN) but they seem to not be mapped to the right code points (probably because the designer's software couldn't handle non-BMP chars).
I'll have to redesign a bunch of stuff though
@Adám are they weirdly thin like the ones from bqn386?
oh, love dotted zero
No, they are a complete rethink of what double-struck means, and I rather like it:
10:39
hmm
I guess it's easy to add the other ones if anything
are they easy to distinguish at small sizes?
@RubenVerg Me too, but the dot might be a little too heavy. Maybe it should be as small as ·
@RubenVerg Probably not. I suppose they'll be distinct by looking un-even in heaviness. Syntax colouring will help too. (In fact, if one has proper syntax colouring for BQN, then they could be identical to the single-stroke letters!)
suppose 𜳖 could be supported with the same system as well
@RubenVerg Is that an outlined A that got mangled?
Yes, that could work by simply making all parts double. Not sure you could distinguish between 𜳞 and 𝕀 but there are plenty of other homoglyphs, so that's probably ok.
yeah it's an outlined A, U+1ccd6
is there small caps support?
No.
And in general, no fancy open type features.
E.g. it might be good to be able to turn off the zero dot.
@RubenVerg Something is funky, though. If I force the BQN website to use the font, the bolds are inconsistent:
I have no idea what happened to N and y here.
11:15
@Adám oh I meant the actual Unicode chars
@RubenVerg Eh, Unicode doesn't have small caps codepoints.
TinyAPL uses smallcaps J and Vyxal 3.5 uses smallcaps R
Oh, that's what you mean. Right, so the font currently doesn't have TinyAPL support, but of course, it can always be extended. I'd prefer PRs over rename-forks, but it is public domain.
@Adám a stylistic set for tildes on and nor and obverse with the tilde in the middle would be nice
@Adám I have a fork just because it's spun of BQN386 and I remember something about not taking prs for other langs
@RubenVerg Sort of. I did experiment with that, and it turns out that the font has such thick stroke that middle just doesn't work.
11:19
if you now support bqn in APL386 I'd love to PR
@Adám ah
@RubenVerg You might want to wait until the designer fixes the last problem and updates the source.
sure yeah
TinyAPL386 has a folder with all the new glyphs - I'll have to store that somewhere else, I guess
11:35
@RubenVerg I don't know enough about the Glyph app (can't try it, since I'm not on a Mac). I don't know about the advantages of separate files and a single large file, nor if they can co-exist. I personally wouldn't mind if the source was split.
I don't use Glyphs either
I just draw separate SVGs in Inkscape and then import them in FontForge
is the new font done in Glyphs?
not sure how I would contribute to it then
Me neither, but maybe there are ways to convert.
12:03
does the font use any Glyphs features FontForge doesn't support?
I don't think so, since it doesn't have any fancy features at all. In fact, I think every glyph is independent, instead of having e.g. á link to a.
12:22
What would be good names to distinguish the numeric vector 13 10 from ⎕UCS 13 10?
Newline and NewlineSequence could be either.
NewlineString for the latter is inaccurate as it is actually a character vector, not a string (and NewlineCharacterVector is just too unwieldy)
 
2 hours later…
14:45
Do you have any idea…
What does the following monadic Boolean function determine when given a Boolean vector argument?
∨/
 
1 hour later…
16:00
@PaulMansour Next build of 20.0 allow ⎕VSET, ⎕VGET, and ⎕NS to take any rank left argument.
16:55
@Adám even GPT-4o-mini can answer that ;D
 
1 hour later…
18:13
@Adám Nice, will check it out.
 
3 hours later…
21:11
∨/ a b c would be a ∨ (b ∨ c)
@Rosario Indeed, but the real question is what it signifies. What question does it answer?

« first day (2945 days earlier)      last day (3 days later) »