Additionally, its seems like print for a matrix is designed to let you
recover the raw matrix somehow, and dump is for humans.
But for strings, dump's output lets you recover it (escaped newlines), and
print is for humans?
On Wednesday, April 30, 2014 3:52:37 PM UTC-4, Iain Dunning wrote:
>
> Hi all,
>
> I'm trying to write some pretty-print code for one of my packages. In
> particular, I want to print tensors, e.g. rand(3,3,3)
> Now the (exported) options seem to be "print" and "dump"
>
> For 2D:
> julia> print(rand(3,3))
> [0.9848410067397786 0.7395088321894974 0.08335407347392443
> 0.9908990998709222 0.6563403470680476 0.9422669325824669
> 0.967619446504451 0.995657061136656 0.14417657902926084]
>
> +? Lots of precision
> - Not aligned
> - Square brackets
>
> julia> dump(rand(3,3))
> Array(Float64,(3,3)) 3x3 Array{Float64,2}:
> 0.60914 0.149033 0.431226
> 0.0413379 0.872331 0.24617
> 0.528959 0.225907 0.938011
>
> + Aligned
> + "Nice" amount of precision
> - Type info
>
> For 3D, the output of print is a bit confusing (just prints the three
> slices - not even clear which slices though), whereas dump prints something
> unambiguous
>
> julia> dump(rand(2,2,2))
> Array(Float64,(2,2,2)) 2x2x2 Array{Float64,3}:
> [:, :, 1] =
> 0.698132 0.646175
> 0.0263156 0.721012
>
> [:, :, 2] =
> 0.687558 0.0205718
> 0.89261 0.383277
>
> Now I was digging around with auto-complete and found the un-exported
> function print_matrix which has the the alignment of dump, and precision of
> print - but only works for <= 2D.
>
> Is there some (reasonable) way to export that nice dump functionality
> without the type info? I tried to trace the dump methods through but got
> lost.
>
