if cond then works

[StefanKarpinski]

julia> if true then
           println("here")
       end
here

Should this be a undefined variable error?

[JeffBezanson]

I feel like this has been brought up a hundred times before but I can't find the references.

[StefanKarpinski]

If you replace true with false I have no problem with this working, but using an undefined variable – even in void context – seems like it should be an error.

[stevengj]

Maybe this is a feature? This way we can make our code more polite by writing

if true please
     println("here")
end

We can even add an INTERCAL-like feature in which a program that does not contain please sufficiently often is rejected by the parser as impolite.

Seriously, though, I'm inclined agree with @StefanKarpinski that using an undefined variable in a void context should be an error. I'm finding a hard time imagining a practical use case for this.

[mbauman]

I was using this for inline comments before we got the #= block syntax:

julia> function foo(bar,baz)
         bar > 1 && (sometimes_frob_the_bar; bar+=1)
         but_always_quux_with_the_baz; baz*bar/2
       end
foo (generic function with 1 method)

julia> foo(2,3)
4.5

(I kid. +1 for errors here, but it's bigger than just if clauses.)

[stevengj]

Right, it's any use of an undefined variable in a void context.