Should overwrite option in prep_conversion regenerate .heudiconv stored files

[alexpron]

Summary

Dear heudiconv experts,

I have kind of a specific use case:

• I have DICOM data that are downloaded in a temporary directory located inside a subdirectory of the a main directory
  where the BIDS conversion is performed with heudiconv and stored as a datalad dataset.
• The final datalad dataset may be copied or moved but data can still be added.
• By default I activated the overwrite option as I wanted the update of the data to be handled by datalad.
• With overwrite and after moving the dataset the conversion on already converted data crashes as edit.txt and heuristic file are identical but filegroup.json contains absolute path that are non longer valid

Hence, should the overwrite option also triggers the full generation of new data including BIDS mapping ?
Or am I doing it the wrong way and should I catch the exceptions before,

Best,
Alexandre

Platform details:

Choose one:

• Local environment

 - OS: Fedora 
 - Python 3.10. 13 (conda virtual environement)

• Container

• Heudiconv version:

1.1.6

[yarikoptic]

I guess there is somewhat two issues

• storage of full path whenever input is given via relative path (we can keep this issue to discuss/fix that)
• inability to completely discard prior conversion mapping. I think it is a duplicate of saving of hidden files that are not updated with changes to heuristic #84 - let's continue there.

[yarikoptic]

also related

• repreprocessing from dicom tarballs with --overwrite would fail since full paths to tempdirs were recorded #103

but there I am not sure we could avoid storing full paths since temporary from extracted from zipball dicoms.

[alexpron]

Thanks @yarikoptic for your detailed answer :)