Error of using Dagger, according to the manual

Specific Domains Julia at Scale
1

I tried Dagger according to the manual and got some errors.

using Dagger
using Distributed

addprocs(3)
@everywhere function f1(v1::Float64, v2::Float64)
    r = 0.
    for i in 1:10^7
        r += sin(v1) + sin(v2)
    end
    return r
end
@everywhere function f2(v::Float64)
    r = 0.
    for i in 1:10^7
        r += cos(v)
    end
    return r
end
T1 = delayed(f1; v1=1., v2=2.)
T2 = delayed(f2; v=1.)(T1)         # error here
T3 = delayed(f1; v1=3., v2=4.)
collect(T1)    # error here

MethodError: no method matching Thunk(::typeof(f2), ::getfield(Dagger, Symbol(“##36#37”)){Base.Iterators.Pairs{Symbol,Float64,Tuple{Symbol,Symbol},NamedTuple{(:v1, :v2),Tuple{Float64,Float64}}},typeof(f1)}; v=1.0)
Closest candidates are:
Thunk(::Any, ::Any…; id, get_result, meta, persist, cache, cache_ref, affinity) at /Users/zhangliye/.julia/packages/Dagger/sdZXi/src/thunk.jl:29 got unsupported keyword argument “v”

Stacktrace:
[1] kwerr(::NamedTuple{(:v,),Tuple{Float64}}, ::Type, ::Function, ::Function) at ./error.jl:125
[2] (::getfield(Core, Symbol(“#kw#Type”)))(::NamedTuple{(:v,),Tuple{Float64}}, ::Type{Thunk}, ::Function, ::Function) at ./none:0
[3] (::getfield(Dagger, Symbol(“##36#37”)){Base.Iterators.Pairs{Symbol,Float64,Tuple{Symbol},NamedTuple{(:v,),Tuple{Float64}}},typeof(f2)})(::Function) at /Users/zhangliye/.julia/packages/Dagger/sdZXi/src/thunk.jl:71
[4] top-level scope at In[32]:17

type or paste code here

Julia version information,

Julia Version 1.1.0
Commit 80516ca202 (2019-01-21 21:24 UTC)
Platform Info:
  OS: macOS (x86_64-apple-darwin14.5.0)
  CPU: Intel(R) Core(TM) i5-8500 CPU @ 3.00GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.1 (ORCJIT, skylake)

Dagger document: Dagger example

2

You’re calling delayed incorrectly. For your first invocation, it would be delayed(f1)(1., 2.). The kwargs to delayed are only used to communicate information to the scheduler, and the call to delayed(f1) returns a function which must then be called with the appropriate arguments to the function it wraps.

@Alessio_Bellisomi is working on a documentation PR to make these points more clear in the README.