D'oh (for myself): this repo (dataset) is pure git, not git-annex
extracted all metadata (without any content - just base core, datalad 0.12.5.dev19), and it went into git.
FWIW some stats
- totals: 5.8GB of .datalad/metadata
.git/objects is 1.6GB
- majority is
ds-* files (~4.9GB). They didn't get .xz'ed
- majority of
ds-* size seems to be due to annex.filename which is a list of lists of filenames.
- those
ds-* compress ~8 times if we do xz them
D'oh (for myself): this repo (dataset) is pure git, not git-annex
extracted all metadata (without any content - just base
core, datalad 0.12.5.dev19), and it went into git.FWIW some stats
.git/objectsis 1.6GBds-*files (~4.9GB). They didn't get .xz'edds-*size seems to be due toannex.filenamewhich is a list of lists of filenames.ds-*compress ~8 times if we do xz them