Fix a few deprecations (JuliaDSP#200)

* Avoid implicit scalar broadcasting in setindex!

* Add `using` for `eig` and `svd` to `Estimation` module

Also drop a spurious `import` statement.

* Move two previously global variables into a testset in LPC tests

src/Filters/filt.jl

@@ -522,7 +522,7 @@ function fftfilt(b::AbstractVector{T}, x::AbstractArray{T},
     # FFT of filter
     filterft = similar(tmp2)
     copyto!(tmp1, b)
-    tmp1[nb+1:end] = zero(T)
+    tmp1[nb+1:end] .= zero(T)
     mul!(filterft, p1, tmp1)
 
     # FFT of chunks
@@ -533,8 +533,8 @@ function fftfilt(b::AbstractVector{T}, x::AbstractArray{T},
             xstart = off - nb + npadbefore + 1
             n = min(nfft - npadbefore, nx - xstart + 1)
 
-            tmp1[1:npadbefore] = zero(T)
-            tmp1[npadbefore+n+1:end] = zero(T)
+            tmp1[1:npadbefore] .= zero(T)
+            tmp1[npadbefore+n+1:end] .= zero(T)
 
             copyto!(tmp1, npadbefore+1, x, colstart+xstart, n)
             mul!(tmp2, p1, tmp1)

src/estimation.jl

@@ -1,5 +1,6 @@
 module Estimation
-import Base: *
+
+using Compat.LinearAlgebra: eig, svd
 
 export esprit
 
@@ -8,10 +9,10 @@ export esprit
 
 ESPRIT [^Roy1986] algorithm for frequency estimation.
 Estimation of Signal Parameters via Rotational Invariance Techniques
-    
+
 Given length N signal "x" that is the sum of p sinusoids of unknown frequencies,
 estimate and return an array of the p frequencies.
-    
+
 # Arguments
 - `x::AbstractArray`: complex length N signal array
 - `M::Integer`: size of correlation matrix, must be <= N.
@@ -21,7 +22,7 @@ estimate and return an array of the p frequencies.
       For faster execution (due to smaller SVD), use small M or small N-M
 - `p::Integer`: number of sinusoids to estimate.
 - `Fs::Float64`: sampling frequency, in Hz.
- 
+
 # Returns
 length p real array of frequencies in units of Hz.
 

test/filter_conversion.jl

@@ -80,8 +80,8 @@ using DSP, Compat, Compat.Test, FilterTestHelpers, Polynomials
 
     # And with only poles (no zeros)
     m_sos_only_poles = copy(m_sos_full)
-    m_sos_only_poles[:, 1:2] = 0
-    m_sos_only_poles[:, 3] = 1
+    m_sos_only_poles[:, 1:2] .= 0
+    m_sos_only_poles[:, 3] .= 1
     @test m_sos_only_poles ≈ sosfilter_to_matrix(convert(SecondOrderSections, ZeroPoleGain(Float64[], p, k)))
 
     # Test that a filter with repeated zeros is handled properly

test/lpc.jl

@@ -1,8 +1,8 @@
 # Filter some noise, try to learn the coefficients
-coeffs = [1, .5, .3, .2]
-x = filt([1], coeffs, randn(50000))
-
 @testset "$method" for method in (LPCBurg(), LPCLevinson())
+    coeffs = [1, .5, .3, .2]
+    x = filt([1], coeffs, randn(50000))
+
     # Analyze the filtered noise, attempt to reconstruct the coefficients above
     ar, e = lpc(x[1000:end], length(coeffs)-1, method)