Is the following a bug in J, or am I missing something? 'ab' ,&(1$~#) 'abcd' works but 'ab' ,&(1"0) 'abcd' gives a length error, even though they should be the same. ATO

'ab' ,&:(1"0) 'abcd' works fwiw, but I don't understand why the parens don't stop the "0 from propagating...

Nm, I see.

@mitchelljohnstone I think it's coming in the next major version

@RubenVerg I was under the same impression, too. Kamila did the (initial?) implementation AFAIU and mentioned in her talk at the User Meeting this year that it might not be in 19.

oh, dyalog 19 is a thing? I'd assumed it would go from 18 to 20 because 19 was taken by the april's fools article

mostly because I'd never seen anyone talk about dyalog 19

@Ven Still not seeing it:

Oh, unevaluated, it just matches the {}s?

@RubenVerg @mitchelljohnstone 20.0.

@Ven No…

@B.Wilson "might not"? Definitely not.

Well, not that simple, but it disregards ()s completely

I guess the role of an element can be modified from outside and change the parse

Can you screenshot an example?

There's nothing to screenshot, I'm observing the same behavior as you. I'm just surprised by the unevaluated case of {(((⍵} printing itself

Oh, well functions (statements, actually) aren't parsed until runtime. It is perfectly allowed to have syntax error in code. Sometimes people even put them in intentionally to halt execution.