What the corpus takes on disk: every file under blobs/, records/ and index/
is listed and its size read (nothing is opened), so on a large corpus this costs
one stat per file — which is why ka stats only does it when asked (--disk).
Platform files (isPlatformFile) are not counted.
What the corpus takes on disk: every file under
blobs/,records/andindex/is listed and its size read (nothing is opened), so on a large corpus this costs onestatper file — which is whyka statsonly does it when asked (--disk). Platform files (isPlatformFile) are not counted.