Unicode comparison operators parsing differs from ASCII counterparts

[carlobaldassi]

In particular, they are not parsed as word-boundaries when tokenizing, i.e. they require spaces around them. For example:

julia> a,b = 3,4;

julia> a!=b
true

julia> a≠b
ERROR: a≠b not defined

julia> a>=b
false

julia> a≥b
ERROR: a≥b not defined

julia> a===b
false

julia> a≡b
ERROR: a≡b not defined