Binary operators and broadcast syntax don't interact well

[fcard]

Currently, it's necessary to wrap binary operators in parentheses to use them with the new broadcast syntax. Otherwise:

julia> *.(2,[1,2,3])
ERROR: syntax: "*" is not a unary operator
 in eval(::Module, ::Any) at ./boot.jl:231
 in macro expansion at ./REPL.jl:92 [inlined]
 in (::Base.REPL.##1#2{Base.REPL.REPLBackend})() at ./event.jl:46

or

julia> +.(1,[1,2,3])
ERROR: syntax: numeric constant "." cannot be implicitly multiplied because it ends with "."
 in eval(::Module, ::Any) at ./boot.jl:231
 in macro expansion at ./REPL.jl:92 [inlined]
 in (::Base.REPL.##1#2{Base.REPL.REPLBackend})() at ./event.jl:46

P.S. Also affects the not nearly as important but still happenable cases of using unicode operators as normal variables:

type GreekTemple
  number_of_columns::Int
end
const ⟰ = GreekTemple(4)

println(⟰.number_of_collumns) 
# My attempt to build a compilable greek city with unicode art is ruined >:

type Sadness
  sadness_rating::Int
end
const >: = Sadness(10000)

println(>:.sadness_rating) # >: >: >: >: >: >:

edit: I was so used calling custom binary operators unicode that I forgot that some of them are ascii...

[JeffBezanson]

The case with the "implicitly multiplied" error looks like a bug; there is no numeric constant present so at the very least it needs a better error.

I don't think we ever intended to allow accessing fields of operators with ., but maybe we can. It's a bit ambiguous with e.g. +.2, which is unary + applied to 0.2.

[fcard]

Well, a.2 is invalid syntax as well, so +.2 could still mean what it does now. And accessing fields of operators already works with (), as in (⟰).number_of_collumns, so that would just make it easier for when it's necessary... although, I don't see myself personally using that, so I am fine either way, just reportin'!
edit: I think I understand better what you said now, that it would be ambiguous to the user, rather than to the compiler. Fair point! Although, if one is redefining common unary operators and accessing their fields, the entire code must already be looking pretty confusing :P

Also, I just noticed that unary operators other than + and - are giving a different error:

julia> !.(1)
ERROR: syntax: invalid identifier name "."
 in eval(::Module, ::Any) at ./boot.jl:231
 in macro expansion at ./REPL.jl:92 [inlined]
 in (::Base.REPL.##1#2{Base.REPL.REPLBackend})() at ./event.jl:46

I am guessing because the parser is trying to parse .(1) separately.
(and - gives numeric constant ".-" instead of just ".")

[stevengj]

Note that this should be mostly addressed in 0.6 by #17623, since you will just do 2 .+ [1,2,3] to get a fused broadcast.

[stevengj]

Closed by 17623.