MethodError: start has no method matching start(::Type{Int64})

[dpo]

The error below is probably due to a very recent change; it didn't happen with 0.4.0-dev+4294 (commit b7d7955):

               _
   _       _ _(_)_     |  A fresh approach to technical computing
  (_)     | (_) (_)    |  Documentation: http:https://docs.julialang.org
   _ _   _| |_  __ _   |  Type "help()" for help.
  | | | | | | |/ _` |  |
  | | |_| | | | (_| |  |  Version 0.4.0-dev+4460 (2015-04-22 21:32 UTC)
 _/ |\__'_|_|_|\__'_|  |  Commit 3bbc5fc (0 days old master)
|__/                   |  x86_64-apple-darwin13.4.0

julia> using DistributedArrays
ERROR: LoadError: MethodError: `start` has no method matching start(::Type{Int64})
 in append_any at base.jl:198
 in include at /Applications/Julia-0.4.0-dev-3bbc5fc0d0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in include_from_node1 at /Applications/Julia-0.4.0-dev-3bbc5fc0d0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in reload_path at /Applications/Julia-0.4.0-dev-3bbc5fc0d0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in _require at /Applications/Julia-0.4.0-dev-3bbc5fc0d0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in require at /Applications/Julia-0.4.0-dev-3bbc5fc0d0.app/Contents/Resources/julia/lib/julia/sys.dylib
while loading /Users/dpo/.julia/v0.4/DistributedArrays/src/DistributedArrays.jl, in expression starting on line 319

[gvdr]

Got the same...

[andreasnoack]

Have you run Pkg.update()?

[gvdr]

Wop, actually error solved after Pkg.checkout("DistributedArrays").

[andreasnoack]

What do you get from versioninfo()?

[gvdr]

Here we are:

Julia Version 0.4.0-dev+4532
Commit 0398bb2 (2015-04-27 16:10 UTC)
Platform Info:
  System: Linux (x86_64-redhat-linux)
  CPU: Intel(R) Xeon(R) CPU E5-2667 0 @ 2.90GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT NO_AFFINITY NEHALEM)
  LAPACK: libopenblas
  LIBM: libopenlibm
  LLVM: libLLVM-3.3

[andreasnoack]

...and Pkg.status()?

[gvdr]

The whole list is:

10 required packages:
 - DistributedArrays             0.1.0+             master
 - GLM                           0.4.6
 - Gadfly                        0.3.11+            master
 - Graphs                        0.5.4
 - IJulia                        0.2.3
 - Optim                         0.4.1
 - PTools                        0.0.0+             master
 - PyCall                        0.8.0+             master
 - PyPlot                        1.5.3
 - RDatasets                     0.1.2
41 additional packages:
 - ArrayViews                    0.6.1
 - BinDeps                       0.3.11
 - Calculus                      0.1.7
 - Codecs                        0.1.4
 - Color                         0.4.4
 - Compat                        0.4.2
 - Compose                       0.3.11
 - Contour                       0.0.6
 - DataArrays                    0.2.13
 - DataFrames                    0.6.4
 - DataStructures                0.3.8
 - Dates                         0.4.4
 - Distances                     0.2.0
 - Distributions                 0.7.1
 - Docile                        0.4.11
 - DualNumbers                   0.1.3
 - FixedPointNumbers             0.0.7
 - GZip                          0.2.15
 - Graphics                      0.1.0
 - Grid                          0.3.8
 - Hexagons                      0.0.3
 - ImmutableArrays               0.0.7
 - Iterators                     0.1.7
 - JSON                          0.4.3
 - KernelDensity                 0.1.0
 - LaTeXStrings                  0.1.2
 - Loess                         0.0.3
 - NaNMath                       0.0.2
 - Nettle                        0.1.8
 - NumericFuns                   0.2.3
 - PDMats                        0.3.3
 - REPLCompletions               0.0.3
 - RandomGraphs                  0.0.0-             master (unregistered)
 - Reexport                      0.0.2
 - SHA                           0.0.4
 - Showoff                       0.0.4
 - SortingAlgorithms             0.0.4
 - StatsBase                     0.6.14
 - URIParser                     0.0.5
 - WoodburyMatrices              0.0.1
 - ZMQ                           0.1.17

[gvdr]

I'm trying to use it, to see how it runs. So far almost so good (almost):

julia> dzeros((100,100), workers()[1:4], [1,4])
ERROR: MethodError: `dzeros` has no method matching dzeros(::Tuple{Int64,Int64}, ::Array{Int64,1}, ::Array{Int64,1})
Closest candidates are:
  dzeros{T}(::Type{T}, ::Tuple{Vararg{Int64}}, ::Any...)
  dzeros(::Tuple{Vararg{Int64}})

[andreasnoack]

Try

dzeros(Float64, (100,100), workers()[1:4], [1,4])

It has just been changed, but I think we should default to Float64.

[gvdr]

Runs smoothly.

[gvdr]

By the side, I'm discovering DistributedArrays after trying to implement something akin to the archetypical problem (mentioned in http:https://julia.readthedocs.org/en/latest/manual/parallel-computing/):

a = zeros(100000)
@parallel for i=1:100000
  a[i] = i
end

and being informed that DArrays is the way to go. So, what would the DArrays approach be?

P.S. Sorry for the off topic, if there is a better place to discuss this please let me know.

[andreasnoack]

I think questions like these fits well on the Julia-users list, but see my reply below anyway.

I think it depends on what you are calculating. If you are running a heavy simulation that you'd like to execute in parallel, while the result can just be processed on the master process, I'd just use something like

@parallel (vcat) for i = 1:100000
1+2+3+i
end

or collect the simulation in a function, say simfun and run

r = pmap(simfun, 1:100000)

In other cases a DArray might be useful. Recently mapslices was extended to work for DArrays so you can do something like

julia> A = drandn((10,10,5000), workers(), [1,1,4]);

julia> mapslices(inv, A, (1,2));

to invert each of the 5000 10x10 matrices in the array.

[gvdr]

👍

Thanks, the mapslices is perfect for my applications, but all the other examples are much useful.

[jakebolewski]

@gvdr I added the missing defaults for dzeros / dones so your above example should work now.

[dpo]

Any idea when the fix will appear in stable?

[jakebolewski]

This package is currently only works for 0.4-dev, so I don't think it makes sense to tag a new release until 0.4 is released.