Add infix "!in" as synonym for ∉ ?

[samoconnor]

I miss not having infix ni as the opposite of in like Tcl has

∉ is nice, but I don't know how to type that.

I end up with a lot of if !(x in v) ... which I'd rather replace with if x ni v ...
Does this seem crazy to people without a Tcl disorder?

Update: waiting for general infix ! per #15353 (comment) .

[nalimilan]

?

[samoconnor]

Sorry @nalimilan I think I pressed enter too soon an submitted a half-typed issue.

[Dawny33]

∉ can be typed by \notin (both in Atom with the latex plugin and the REPL)

[samoconnor]

HI @Dawny33, I'm not sure what Atom is. I mostly only ever type code into Vim. I'm sure there is a way to type ∉ in Vim, but I can only fit a finite number of Vim shortcuts in my brain.

I think it's reasonable to expect that non-ascii operators have an ascii-typeable equivalent.

[Dawny33]

@samoconnor Yeah, the request is quite reasonable 👍

Just trying to point out the latex shortcut for quick reference!

[nalimilan]

Does this seem crazy to people without a Tcl disorder?

Yes, I must confess I'm in that camp. :-) Using if/fi and things like that always sounded weird to me. We may as well use out (just joking).

[samoconnor]

ni is nice because it is reverse("in") and also "not in" and also because Ni!

But if ni is too cute 🐶, then !in or not in or notin would be fine too...

[tkelman]

If we really need an ascii infix not-in operator (is !(a in b) not sufficient for most purposes?) then I'd rather have it spelled notin than ni.

[Dawny33]

@tkelman +1

notin plays well with the latex shortcut \notin, so it avoids unnecessary confusion. And yeah, we need a not in operator, as it is a bit of a frequent usage.

[KristofferC]

Ni! Ni! Ni!

[samoconnor]

@KristofferC Icky Icky Icky Ptang

[nalimilan]

Indeed, the movie reference is the only possible justification of Tcl's choice.

[samoconnor]

Adding notin(x, itr) = !in(x,itr) to reduce.jl is easy enough.
What has to be changed to make it work as an infix operator?

[eschnett]

Clearly, ni should be equivalent to \ni, which is∋-- the left-right swapped version ofin`.

[vchuravy]

@samoconnor You might want to look at https://github.com/JuliaLang/julia-vim AFAIK it can replace \notin with the appropriate unicode symbol

[StefanKarpinski]

Perhaps parsing !in as infix would be better, as in a !in b.

[kmsquire]

+1 for !in as an infix operator.

[JeffBezanson]

One reason to prefer notin is that it's compatible with allowing all functions to use infix syntax, while !in would continue to be a special case even if we made that change.

[StefanKarpinski]

If a foo b desugars to foo(a,b) then a !foo b could generically desugar to !foo(a,b). On the other hand, I find this particular syntax issue to be mild enough that it can wait.

[nalimilan]

Could be useful, but we'll have to wait until more infix operators are supported to really give !in that general meaning. Until then, it will only be a weird exception.

[samoconnor]

If it is agreed that a general infix ! and !in can wait, then notin seems to be broadly agreeable.
Would a PR for notin be welcome?

[StefanKarpinski]

I'm against it because then we'd have to deprecate notin when !in happens. In the meantime, writing !(a in b) is not so awful.

[samoconnor]

Fair enough.

[KristofferC]

Now have !in

[DNF2]

Does it also solve #25512, or does it only work for in?

[nalimilan]

But x !in y doesn't work, right? Yet that's what this issue is about.

[KristofferC]

Yeah, sorry, I missed the actual point here.

[Moelf]

I think people are upset that you have to write !in(2, [2,3]) but can't write 2 !in [2,3]

[StefanKarpinski]

Allowing ! before infix word operators (in, isa) seems like it would ok to me. However it's slightly dangerous territory since by analogy if a !in b means !(a in b) then one would expect that a !== b to mean !(a == b) but it actually means !(a === b).

[Moelf]

maybe disallow !== say ambiguity error?

[yurivish]

You also can't broadcast in in infix form, and I think it would be nice if you were able to do that.

julia> in.(1, [[1, 2], [3, 4]])
2-element BitArray{1}:
  true
 false

julia> 1 .∈ [[1, 2], [3, 4]]
2-element BitArray{1}:
  true
 false

julia> 1 .in [[1, 2], [3, 4]]
ERROR: syntax: space before "." not allowed in "1 ."

[StefanKarpinski]

maybe disallow !== say ambiguity error?

Yeah, to hell with no breaking changes 😂

[Moelf]

oh... right !== is already there sorry

[nvhoang3110]

Please is there any news regarding to the issue ? Some think like not in just like in Python would be nice.

[Moelf]

I feel like the best we can do is special case notin as an infix thing, similar to isa, since it's very clear that general !<infix op> can't happen