Consistent array slicing with Base

[IlyaOrson]

Maybe the getindex() behaviour should be the same on the data stored in HDF5 and the one stored in a normal array in julia:

julia> h5open("test.h5","w") do file
           rows, columns = 3, 12
           data = zeros(rows, columns)
           group = g_create(file, "group")
           data_hdf5 = d_create(group, "data", datatype(Float64), dataspace(rows, columns), "chunk", (rows, 1))
           for i = 1:columns
               vec = rand(3)
               data[:,i] = vec
               data_hdf5[:,i] = vec
           end
           typeof(data[:,1]), typeof(data_hdf5[:,1])
       end
(Array{Float64,1}, Array{Float64,2})

[IlyaOrson]

A related example where the normal syntax that works on a julia array fails with HDF5 data set:

julia> h5open("test.h5", "w") do file
       b = d_create(file, "b", Int, ((1000,50),(-1,-1)), "chunk", (1000,1)) 
       b[:,:] = collect(1:length(b)) # This works
       b[:] = collect(1:length(b)) # This fails
       end
ERROR: number of elements in range and length of array must be equal
Stacktrace:
 [1] _setindex!(::HDF5.HDF5Dataset, ::Type{T} where T, ::Array{Int64,1}, ::UnitRange{Int64}, ::Vararg{UnitRange{Int64},N} where N) at /home/ilya/.julia/v0.6/HDF5/src/HDF5.jl:1644
 [2] setindex!(::HDF5.HDF5Dataset, ::Array{Int64,1}, ::UnitRange{Int64}) at /home/ilya/.julia/v0.6/HDF5/src/HDF5.jl:1633
 [3] (::##35#36)(::HDF5.HDF5File) at ./REPL[39]:4
 [4] h5open(::##35#36, ::String, ::String) at /home/ilya/.julia/v0.6/HDF5/src/HDF5.jl:590

My version info:

julia> versioninfo()
Julia Version 0.6.1-pre.0
Commit dcf39a1dda (2017-06-19 13:06 UTC)
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: Intel(R) Core(TM) i7-4510U CPU @ 2.00GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Haswell)
  LAPACK: libopenblas64_
  LIBM: libopenlibm
  LLVM: libLLVM-3.9.1 (ORCJIT, haswell)

julia> Pkg.status("HDF5")
 - HDF5                          0.8.1+             master

[kleinhenz]

the first example works as expected since #603. The second example still doesn't work because hyperslab doesn't support 1D indexing of multidimensional arrays. This seems definitely doable to implement if we want to.