BUG in a42a?

[sglyon]

Here is a (partial) traceback:

ERROR: UndefVarError: e not defined
 in a42a at /Users/sglyon/.julia/v0.4/Roots/src/fzero.jl:165
 in fzero at /Users/sglyon/.julia/v0.4/Roots/src/Roots.jl:114

I've looked over and over this function and cannot figure out why this is happening. The && should short-circuit so the distinct(f, a, b, d, e) should only run after the first iteration has completed. Then e is defined towards the end of every iteration, so by the time this ever executes e should be in scope...

EDIT: link to the line in question

[yuyichao]

I think the scope of the variable does not escape the iteration unless you specify so.

julia> function k1()
           for i in 1:3
               i == 1 || println(f)
               f = 1
           end
       end
k1 (generic function with 1 method)

julia> function k2()
           local f
           for i in 1:3
               i == 1 || println(f)
               f = 1
           end
       end
k2 (generic function with 1 method)

julia> k1()
ERROR: UndefVarError: f not defined
 in k1 at ./none:3

julia> k2()
1
1

[sglyon]

Hmm that's subtle. I didn't know that.

[yuyichao]

Yeah. The document of the scope definitely needs more comprehensive documentation JuliaLang/julia#9955

[jverzani]

Hopefully this f5f3789 fixes the issue. (I couldn't reproduce the error to be sure.)