Should `?` not be allowed as an identifier name?

[carlobaldassi]

Given its role in ternary expressions (and perhaps in more upcoming syntactic constructs?), isn't letting ? be used as a variable name allowing too much room for potential confusion?

julia>  ? = true
true

julia>  ??!?:?
false

julia>  !??!?:!??!?:?
true

julia>  !???:!?:?
false:true

julia> (x->(?->?(?))(?->x(x->?(?)(x))))(x->(?->(?==0?1:(?*x(?-1))))).(?:?<<?<<?)
4-element Array{Int64,1}:
  1
  2
  6
 24

(also isn't it amazing that these even parse?) Anyway, I suppose, unless one really wants to use ? as a function or macro name, that this possibility could be deprecated?

[tkelman]

Duplicate of #6286 ? Now that @unix? macros are deprecated we can hopefully tweak the parsing here.

[andyferris]

Haha I discovered this recently and kind of trolled myself by creating this (an attempt originally inspired by trying to have a more Swift-like syntax for nullables).

Do you think ? could parse successfully as a prefix unary operator without screwing with the ternary?

[nalimilan]

Note that in Swift and C#, the ? denoting nullable/optional is a suffix.