mechababs presentation

[yarikoptic]

• babs (including dashboard) - 2-3 slides
  • "user does not do datalad"
  • built atop datalad run + save
• mechababs
  • "function in life" (1,000s of datasets) (easy bootstrap of babs)
  • approach
    • "user does not do datalad" (good to adopt good approaches)
    • mechababs iterate (so maintains a state machine)
    • ideal to aim: mechababs datalad run on higher level
  • summary of contributions back to babs:
    • ...
  • observed cons/problems
    • datalad run requires clean state -- "LONG TIME TO ASSESS"
      • future TODO: ephemeral local clones, only studies "in work" installed -- the rest in remotes
• mechababs "study-first" edition: Study-first direction — docs RFC, implementation to follow (draft) #101
  • stronger relation to BIBS study
  • better ...
• what/how could we adopt from nipoppy
  • boutiques
  • how will we do BEP028 PROV (@yarikoptic already mumbled about support Python library)
    • mechababs ATM produces some tentative record
    • how to make it (re)actionable

  • how to converge on BIDS study for all?