Fourier with BigFloat (ERROR: interlace not defined)

Hi,

When trying to make Fourier approximations work with BigFloat coefficients instead of Float64 ones, eg in the following minimal example:

using ApproxFun

# FS = Fourier(0..2π)
FS = Fourier(BigFloat(0)..2*BigFloat(π))
f = Fun(exp ∘ cos, FS)

I get the following error trace (which does not occur with Float64 instead of BigFloat):

UndefVarError: interlace not defined
in top-level scope at these/hilbert16/src/abintvalid/test-interlace.jl:6
in Fun at ApproxFunBase/jsbXG/src/constructors.jl:177
in default_Fun at ApproxFunBase/jsbXG/src/constructors.jl:121
in default_Fun at ApproxFunBase/jsbXG/src/constructors.jl:72
in default_Fun at ApproxFunBase/jsbXG/src/constructors.jl:57
in default_Fun at ApproxFunBase/jsbXG/src/constructors.jl:44
in transform at ApproxFunFourier/xV5Mx/src/ApproxFunFourier.jl:508 
in * at ApproxFun/GOMNa/src/Extras/fftGeneric.jl:119

The packages I'm using:

ApproxFun v0.12.5
ApproxFunBase v0.4.3
ApproxFunFourier v0.2.9

If it helps, an easy patch that I've been using so far is to replace the two occurences of "interlace" in lines 119 and 120 in
ApproxFun.jl/src/Extras/fftGeneric.jl by "ApproxFunBase.interlace".

Thanks in advance :)
Florent

The failure is reported in ApproxFun.jl/src/Extras/fftGeneric.jl at lines 119-120; start by tracing the BigFloat Fourier path from the minimal example and checking how interlace is resolved. Done means the example constructs the Fourier Fun with BigFloat coefficients without UndefVarError, while Float64 behavior remains intact.