The new version looks great. Please merge it into the PR.
I have only one small request, but I somehow cannot comment on the commit 🙄
mask_dataset = results.get("mask_ds",…
The current additional mask logic is flat-out 💩. No question there.
I have it specifically on my list to merge it with the OTHER additional mask that we have in the DAQ: https://github.com/pau…
Ah! Now I get the idea!
I think the canonical way of doing this is: https://docs.h5py.org/en/stable/high/dataset.html#h5py.Dataset.astype
I would guess that your version is no extra copy…
Ah, yes, that was clear :)
The reason I got confused is that for such a case, I would use np.array (i.e., the definite-copy version). np.asarray is the copy-if-needed version that you'd…
Please change this to:
Just for my eye, I'd change this to:
Same as above:
Isn't this the same as:
Umm, this is odd 😆 Why not:
I have not seen this before. Is there an advantage over:
Small useless optimization, but also nicer to read: You can drop the [] and use a generator expression.
You are changing the mask in place.
Same as above: drop [].
Why this maybe-copy?
I kinda feel this would be nicer to read: