Closed oesteban closed 5 years ago
Yeah, I'm not really sure how things work without datalad. I guess you could include a repo revision hash that also goes into S3, make a best-effort check for the single S3 file and re-download if the hash has changed?
For S3-backed systems, this requires 1) identifying that this is an update; 2) unzip new files in the skeleton.
For Datalad-backed systems, this requires 1) identifying that this is an update; 2)
datalad update --merge -r
.