Can => be put even lower in operator precedence?

[IainNZ]

Just got caught out by combining => with ? :, e.g.

julia> Dict(1 => true ? println("foo") : println("bar"))
ERROR: TypeError: non-boolean (Pair{Int64,Bool}) used in boolean context

Except the situation I was in was

julia> x = Dict(:foo => 2)
Dict{Symbol,Int64} with 1 entry:
  :foo => 2

julia> Dict(:fizz => 1 == 1 ? x[:foo] : x[:bar])
ERROR: KeyError: bar not found
 in getindex at dict.jl:718

Which was very confusing, as I was wondering why would it be evaluating x[:bar]

[rened]

The precedence was recently increased (#12285), actually. See also Jeff's comment: #12285 (comment)

I'd also like to see it more on the level of =, that's what the => feels most like for me.

[IainNZ]

Ah I see, essentially the same as what you said! Agreed that it feels like it should be at the level of =, its got almost the same semantics except I guess in that one case that @JeffBezanson was showing with the in thing. I guess this is a wont-fix then :(

[rened]

@IainNZ - just to make sure you see the ongoing discussion on #12285 (comment), perhaps this is not a "won't fix" any more

[IainNZ]

👍