Currently, we use .tar.xz for GHC, Cabal, and sandbox archives, and .tar.gz for application archives.
Instead, we could initially use .tar.gz for all archives, and subsequently recompress everything as .tar.xz (with massive XZ_OPTS), using an independent agent running as an external service.
This requires extending storage functions to look for the requested archives in more than one format.
Currently, we use
.tar.xz
for GHC, Cabal, and sandbox archives, and.tar.gz
for application archives.Instead, we could initially use
.tar.gz
for all archives, and subsequently recompress everything as.tar.xz
(with massiveXZ_OPTS
), using an independent agent running as an external service.This requires extending storage functions to look for the requested archives in more than one format.