Error when running the example in the readme

[kskyten]

Running the following with the current registry version of Soss and on the master branch:

using Soss, Random

m = @model X begin
    β ~ Normal() |> iid(size(X,2))
    y ~ For(eachrow(X)) do x
        Normal(x' * β, 1)
    end
end;

Random.seed!(3)
X = randn(6,2)
rand(m(X=X))

gives me the following error:

ERROR: UndefVarError: interpret not defined
Stacktrace:
 [1] type2model(::Type{Model{NamedTuple{(:X,),T} where T<:Tuple,TypeEncoding(begin
    β ~ Normal() |> iid(size(X, 2))
    y ~ For(eachrow(X)) do x
            Normal(x' * β, 1)
        end
end),TypeEncoding(Main)}}) at ~/.local/share/julia/packages/Soss/g2PnG/src/core/model.jl:35
 [2] macro expansion at ~/.local/share/julia/packages/Soss/g2PnG/src/primitives/rand.jl:16 [inlined]
 [3] #s54#71(::Any, ::Any, ::Any) at ~/.local/share/julia/packages/GeneralizedGenerated/NDqgV/src/closure_conv.jl:121
 [4] (::Core.GeneratedFunctionStub)(::Any, ::Vararg{Any,N} where N) at ./boot.jl:524
 [5] rand(::Soss.JointDistribution{NamedTuple{(:X,),Tuple{Array{Float64,2}}},NamedTuple{(:X,),T} where T<:Tuple,TypeEncoding(begin
    β ~ Normal() |> iid(size(X, 2))
    y ~ For(eachrow(X)) do x
            Normal(x' * β, 1)
        end
end),TypeEncoding(Main)}) at ~/.local/share/julia/packages/Soss/g2PnG/src/primitives/rand.jl:8
 [6] top-level scope at REPL[14]:1

[kskyten]

The example works in the dev branch.

[ghost]

Are you on 1.2.0?

julia> using Soss, Random;

julia> m = @model X begin
           β ~ Normal() |> iid(size(X,2))
           y ~ For(eachrow(X)) do x
               Normal(x' * β, 1)
           end
       end;

julia> Random.seed!(3);

julia> X = randn(6,2);

julia> rand(m(X=X));

(v1.2) pkg> status Soss
    Status `~/.julia/environments/v1.2/Project.toml`
  [0bf59076] AdvancedHMC v0.2.13
  [31c24e10] Distributions v0.21.8
  [bbc10e6e] DynamicHMC v2.1.1
  [1a297f60] FillArrays v0.7.4
  [f6369f11] ForwardDiff v0.10.6
  [6fdf6af0] LogDensityProblems v0.9.2
  [d8e11817] MLStyle v0.3.1
  [91a5bcdd] Plots v0.27.0
  [8ce77f84] Soss v0.7.0 #master (https://github.com/cscherrer/Soss.jl.git)
  [4c63d2b9] StatsFuns v0.9.0
  [84d833dd] TransformVariables v0.3.8
  [de0858da] Printf 
  [9a3f8284] Random 
  [10745b16] Statistics

[kskyten]

Yes, I'm using 1.2.0.

[cscherrer]

Thanks @kskyten, I think this is due to a change in GeneralizedGenerated 0.2; I didn't have version upper bounds (oops).

Works in dev (as you say), should also work to downgrade GeneralizedGenerated

[cscherrer]

Fixed in v0.8:

julia> using Soss, Random

julia> m = @model X begin
                  β ~ Normal() |> iid(size(X,2))
                  y ~ For(eachrow(X)) do x
                      Normal(x' * β, 1)
                  end
              end;

julia> Random.seed!(3);

julia> X = randn(6,2)
6×2 Array{Float64,2}:
  1.19156    0.100793  
 -2.51973   -0.00197414
  2.07481    1.00879   
 -0.97325    0.844223  
 -0.101607   1.15807   
 -1.54251   -0.475159  

julia> rand(m(X=X))
(X = [1.1915557734285787 0.10079289135480324; -2.5197330871745263 -0.0019741367391015213; … ; -0.1016067940589428 1.158074626662026; -1.5425131978228126 -0.47515878362112707], β = [0.07187269298745927, -0.5128103336795292], y = [0.10079289135480324, -2.5197330871745263, 2.0748097755419757, 0.8442227439533416, 1.158074626662026, -0.47515878362112707])