« first day (194 days earlier)      last day (1070 days later) » 

3:49 AM
I feel like ǀ is common enough to warrant a single byte equivalent
 
 
12 hours later…
3:43 PM
I had a neat idea for a hyper, so I implemented it! is “sort by length” in my latest pull request.
 
Saves 2 bytes over L€Ụị, and I guess doesn't get confused if you need to do something after it, sounds pretty good.
Also, aren't they all quicks now? I thought hypers were being converted
 
Oh, I suppose the conversion isn’t done yet.
 
Soon™ I guess ;)
 
Hi, @Dennis
 
4:00 PM
Hello!
 
I think my observation about ǀ being common was maybe just clustering bias? I searched for Jelly answers on the site and all the ones that contain it are mine!
 
Haha.
\nÇ€ is common, yeah, but it can oftentimes get shortened to µ€.
 
However: I’m not totally sure about this, but I think µfµ€ is usually equivalent to f¶Ç€
Right
And that’s like, the worst case, when f is more than two links. But f¶Ƈ would be one byte shorter!
And I do think the pattern of “define some function, then map with it in the next one” is very common.
Do you have plans for ƁƇƊƑƓƘⱮƝƤƬƲȤɓƈɗƒɠɦƙɱɲƥʠɼʂƭʋȥ? I don’t think any of them is used, currently
Also, I don’t understand ØV
 
@Lynn Those were meant to be niladic links. Input etc.
 
Is “ṘV” common/special?
 
4:11 PM
There are too many of them though. I'll kick them when I revamp the code page.
There are a lot of backwards incompatible changes I want to make, but I don't want to break existing answers. :/
 
Ahh, yeah. Maybe it’s time to fork into Jelly 2, at some point :)
2
 
Really cheaty and probably won't stay, but there's just no sane way of nesting string delimiters for now.
 
Hahaha
That is quite cheaty
 
4:25 PM
I had ideas for some more quicks based on common constructs of the form f g f⁻¹ for some bijection f
For example, “split on newlines → do a thing → join by newlines” is common, so a “with lines” quick might be nice
 
If I implement inverses properly, J's dual would be a great addition.
 
D → do a thing → ” is another example
 
The next quick I implement with definitely be x mdQ y -> m(x) d m(y). Can't count the times I've needed that.
Speaking of quicks, I'll take a look at your sort-by later. I want to turn it into a quick and let it take niladic and dyadic links as well.
 
Maybe when given a dyadic link, it should use that as a cmp function?
 
Ah, that's an interesting alternative to what I had in mind.
 
4:31 PM
Not sure about niladic ones
It’s hard to think of cmp functions that aren’t just key(r) - key(l) anyway ^^
 
Certainly. Niladic could be sort-as. For example, “abc”1,2,1Þ would result in “acb”.
 
Ah, I like that
 
Although there's a decent chance sort-as would be more useful as an atom.
 
Hm, yeah
 
5:06 PM
“you have three options: …” :)
 
Well, you do have three options. You just happen to also have another option... ;)
 
 
6 hours later…
11:03 PM
I made a little Jelly REPL! We probably don’t need one, because TIO exists. But it’s nice.
 
Actually, it'd probably be useful anyway, since functions are definable
So it'd be good for testing those :)
 
Python sure makes creating REPLs very simple…
    class JellyREPL(code.InteractiveConsole):
        def runsource(self, code, filename='<input>', symbol='single'):
            code = ''.join(char for char in code if char in jelly.code_page)
            jelly.output(jelly.jelly_eval(code, args[:2]), '\n')

    code.InteractiveConsole.interact(JellyREPL(), 'Jelly')
That’s all it does, for now.
So you can’t reuse functions, sadly
Even if I did implement that, it’d be strange, because like, Ç would always refer to the previous line? So you’d have to use that one other quick
 
Could be REPL-only syntax that lets you define your own functions and name them just for testing purposes or something :P
 

« first day (194 days earlier)      last day (1070 days later) »