Currently in restore_version we read and write the entire data to a new version. There was a more efficient implementation of this but that was reverted as it might cause corruptions.
What we want is to have a version that just reads and writes uncompressed chunks to save up on memory blowing up due to the new bytearray + collecting uncompressed data chunks which we ran into.
Currently in restore_version we read and write the entire data to a new version. There was a more efficient implementation of this but that was reverted as it might cause corruptions.
What we want is to have a version that just reads and writes uncompressed chunks to save up on memory blowing up due to the new bytearray + collecting uncompressed data chunks which we ran into.