Debug raw dataset view for two apparently identical datasets

[ehennestad]

Hi,

I am debugging something for matNWB and I need some guidance to understand how neurosift displays some very specific components.

I am looking at these two files: 1) https://neurosift.app/?p=/nwb&url=https://api.dandiarchive.org/api/assets/f02db27e-82eb-41dd-865a-a08bb41491da/download/&dandisetId=000728&dandisetVersion=draft

2) https://neurosift.app/?p=/nwb&url=https://data-proxy.ebrains.eu/api/v1/buckets/d-100b4026-b64e-441d-ba55-9bf2c0213bc4/ophys_tutorial_14.nwb

Specifically, the view I do not understand is this: Raw -> Processing/ophys/ImageSegmentation/PlaneSegmentation/pixel_masks

and I attached two screenshots below to illustrate.

I have looked at the data with a h5 display tool and also in the debug console from neurosift and as far as I can understand, the way the data is saved in h5 should be identical, i.e using the same dataset names, data types etc. However, for the 1st file, the dataset info shows |O and later _COMPOUND_DTYPE: [[x, uint32], [y, uint32], [weight, float32]] whereas for the second file the dataset info shows as object Object. Would you intuitively know if there could be anything different between these two datasets and/or also point me to the relevant pieces of code to understand the situation better.

File 1:

File 2:

[ehennestad]

For more context, see https://github.com/NeurodataWithoutBorders/matnwb/issues/572

[magland]

Hi @ehennestad . The difference is that the first example (since it is public dandi dataset) uses lindi for faster loading. If you don't want to allow lindi loading you can set the lindi=0 query parameter in the URL

See https://github.com/neurodatawithoutborders/lindi