HDF5: varying size dataset

I’m struggle with using HDF5.

I’m trying to append data in .h5 file.
So, I wanna make a dataset whose prescribed size is unlimited (on the first axis, specifically).
For example, receiving data with dims (5,6), then append the existing array with size (n, 5, 6) to be (n+1, 5, 6).

I’ve read this docs, but I don’t understand what I have to do for that.

Here is an example code:

function test_hdf5()
    # using HDF5 (done at the beginning of this file)

    path = "data/tmp.h5"
    name = "asdf"
    val = zeros(1, 5, 6)

    h5open(path, "w") do dummy
    end  # generate `data/tmp.h5`
    h5open(path, "r+") do h5file
        dset = d_create(
                        h5file,
                        name,
                        Float64,
                        (size(val), (-1, size(val)[2:end]...)),  # (1)
                        "chuck",  # (2)
                        size(val),  # (3)
                       )
    end
end

1. do I have to write dims and max_dims (see (1))? Can’t I just write information like (typemax(Int64), 5, 6)?
2. Do I have to add “chuck”? and size(val)?
3. To do what I need, what should be changed in the example code?

Sorry for my poor understand in use of HDF5.

1 post - 1 participant

Read full topic