How to create Compound data with fixed shape?

[nikich340]

I am getting out of my mind trying to create Dataset with type Compound, Shape = N and fields.
That's the original file which I try to recreate from my code:

That's how values look like:

But all I get always is "variable-length" scalar Compound:

I tried to check data types with dataset.Read<Dictionary<string, object>>() on "xformframes" dataset var but it crashes with "Exception: The total file selection element count does not match the total memory selection element count.".
Actually, I'm out of ideas how to archive that, the documentation has no any similar examples.

[Apollo3zehn]

Hi, could you please attach a small sample that shows how you try to receate the expected structure?

[nikich340]

I tried several variants because there are not any examples for pureHDF. That's my last try:

re_anim_clip[boneNames[i]] = new H5Group()
{
    ["xformframes"] = new Dictionary<String, float[][]>()
    {
        { "pos.x", new float[][] {new float[]{0}, new float[]{1}, new float[] {2} } },
        { "pos.y", new float[][] {new float[]{0}, new float[]{1}, new float[] {2} } },
    }
};

I switched to h5py where I finally managed to recreate the structure with this code:

anim_node_data = []
for frame in range(numFrames):
    # should be a tuple, not array!
    anim_node_data.append((
        0.0, 0.0, 0.0,
        0.0, 0.0, 0.0, 1.0,
        1.0, 1.0, 1.0,
        float(frame),
    ))
headers = ['pos.x', 'pos.y', 'pos.z', 'rot.x', 'rot.y', 'rot.z', 'rot.w', 'scale.x', 'scale.y', 'scale.z',
           'time']
ds_dt = numpy.dtype({
    'names': headers,
    'formats': ['float32'] * len(headers),
})
re_anim_node_data = numpy.array(anim_node_data, dtype=ds_dt)
bone_group.create_dataset('xformframes', shape=(numFrames), data=re_anim_node_data)