@cmaumet would you agree on completely removing the software provenance objects in this example ?
I'm afraid it creates an ambiguity as with the current version we are not able to describe if the software was solely responsible for generating the data or if a human intervened in the process.

@cmaumet would you agree on completely removing the software provenance objects in this example ? I'm afraid it creates an ambiguity as with the current version we are not able to describe if the software was solely responsible for generating the data or if a human intervened in the process.

Yes @bclenet I concur 💯

@cmaumet following our discussion from yesterday, I created a provEntity for the source file in this example because it is not in the dataset where the provenance is described (source file is in the sourcedata/raw dataset, provenance is in the derivatives/seg dataset). Nothing in the current version of the specification PR seems to encourage or discourage this. We only have:

Each file with a ent suffix is a JSON file describing provEntities.
These files SHOULD not contain provEntities describing data files that are available in the dataset. Use sidecar JSON files instead for this purpose (see Provenance of a BIDS file).

Which we could change to :

These files SHOULD not contain provEntities describing data files available in BIDS datasets.

But in some (most ?) cases we don't have access to sidecar JSONs for files outside the dataset. This causes a problem for storing Digest and Type metadata.

@bclenet - I think the hope was that we could declare the subdataset in dataset links and use an identifier based on that link (which therefore would make it possible to refer to BIDS files in other datasets). Does that make sense? If not let's go through the example together!

@cmaumet yes, we can refer to a BIDS file in another dataset using its BIDS URI only. But we cannot provide further provenance metadata for this file unless we create a provEntity in a provenance file for it.