update/release: reproducibility test without xz

in testing, between two linux distros, i got different src
tarball hashes, but same hashes when run twice on each

i'm using pretty solid options in tar, and riku came up
with the idea that maybe xz might be causing issue.

i'm ruling this out by running the release script with
this patch, on the same two systems, which will create
a .tar file instead, and i'll compare hashes.

Signed-off-by: Leah Rowe <leah@libreboot.org>
This commit is contained in:
Leah Rowe
2023-10-17 16:24:41 +01:00
parent 4cdf60e60a
commit 9acb6a5339
+2 -2
View File
@@ -232,11 +232,11 @@ mktarball()
[ "${2%/*}" = "${2}" ] || x_ mkdir -p "${2%/*}" [ "${2%/*}" = "${2}" ] || x_ mkdir -p "${2%/*}"
if [ "${tar_implementation% *}" = "tar (GNU tar)" ]; then if [ "${tar_implementation% *}" = "tar (GNU tar)" ]; then
tar --sort=name --owner=root:0 --group=root:0 \ tar --sort=name --owner=root:0 --group=root:0 \
--mtime="UTC 2023-10-14" -c "${1}" | xz -T0 -9e > "${2}" || \ --mtime="UTC 2023-10-14" -c "${1}" > "${2}" || \
err "mktarball 1, ${1}" err "mktarball 1, ${1}"
else else
# TODO: reproducible tarballs on non-GNU systems # TODO: reproducible tarballs on non-GNU systems
tar -c "${1}" | xz -T0 -9e > "${2}" || err "mktarball 2, ${1}" tar -c "${1}" > "${2}" || err "mktarball 2, ${1}"
fi fi
( (
[ "${2%/*}" != "${2}" ] && x_ cd "${2%/*}" [ "${2%/*}" != "${2}" ] && x_ cd "${2%/*}"