summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorIan Jackson <ijackson@chiark.greenend.org.uk>2024-02-09 11:58:45 +0000
committerIan Jackson <ijackson@chiark.greenend.org.uk>2024-02-09 12:00:00 +0000
commit2ff86b9e450de88ae80b8230f191d5e024adef54 (patch)
treee4daea31f1b121922edd05dcd9254e50532f0ed4
parent06c5ba727347fc492530e0dc0606164983d4a316 (diff)
tests: Note results of benchmarking of the gzip -1 flag on artifacts
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
-rw-r--r--tests/lib6
1 files changed, 6 insertions, 0 deletions
diff --git a/tests/lib b/tests/lib
index 883912c..2731f1e 100644
--- a/tests/lib
+++ b/tests/lib
@@ -507,6 +507,12 @@ t-save-artifacts () {
artifacts="$AUTOPKGTEST_ARTIFACTS"
if [ x"$artifacts" = x ]; then return; fi
if [ x"tmp" = x ]; then return; fi
+ # Empirically in a non-autopkgtest run with AUTOPKGTEST_ARTIFACTS
+ # this `-1` flag to gzip decreased the runtime from 585s (2558s CPU)
+ # to 609s (2730s CPU), a decrease of 5-7%. It increased the
+ # total size of the artifacts from 24220k to 26092k, ie +8%,
+ # but in a real autopkgtest run there are many bigger things
+ # alongside AUTOPKGTEST_ARTIFACTS put there by autopkgtest itself.
GZIP=-1v tar -C "$tmp" -zc -f "$artifacts/${0##*/}.tar.gz" \
--exclude=\*.tar .
local logfile="$DGIT_TEST_REDIRECTED_LOG"