git/t/perf/perf-lib.sh

204 lines
5.5 KiB
Bash
Raw Normal View History

Introduce a performance testing framework This introduces a performance testing framework under t/perf/. It tries to be as close to the test-lib.sh infrastructure as possible, and thus should be easy to get used to for git developers. The following points were considered for the implementation: 1. You usually want to compare arbitrary revisions/build trees against each other. They may not have the performance test under consideration, or even the perf-lib.sh infrastructure. To cope with this, the 'run' script lets you specify arbitrary build dirs and revisions. It even automatically builds the revisions if it doesn't have them at hand yet. 2. Usually you would not want to run all tests. It would take too long anyway. The 'run' script lets you specify which tests to run; or you can also do it manually. There is a Makefile for discoverability and 'make clean', but it is not meant for real-world use. 3. Creating test repos from scratch in every test is extremely time-consuming, and shipping or downloading such large/weird repos is out of the question. We leave this decision to the user. Two different sizes of test repos can be configured, and the scripts just copy one or more of those (using hardlinks for the object store). By default it tries to use the build tree's git.git repository. This is fairly fast and versatile. Using a copy instead of a clone preserves many properties that the user may want to test for, such as lots of loose objects, unpacked refs, etc. Signed-off-by: Thomas Rast <trast@student.ethz.ch> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-02-17 18:25:09 +08:00
#!/bin/sh
#
# Copyright (c) 2011 Thomas Rast
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see http://www.gnu.org/licenses/ .
# do the --tee work early; it otherwise confuses our careful
# GIT_BUILD_DIR mangling
case "$GIT_TEST_TEE_STARTED, $* " in
done,*)
# do not redirect again
;;
*' --tee '*|*' --va'*)
mkdir -p test-results
BASE=test-results/$(basename "$0" .sh)
(GIT_TEST_TEE_STARTED=done ${SHELL-sh} "$0" "$@" 2>&1;
echo $? > $BASE.exit) | tee $BASE.out
test "$(cat $BASE.exit)" = 0
exit
;;
esac
TEST_DIRECTORY=$(pwd)/..
TEST_OUTPUT_DIRECTORY=$(pwd)
if test -z "$GIT_TEST_INSTALLED"; then
perf_results_prefix=
else
perf_results_prefix=$(printf "%s" "${GIT_TEST_INSTALLED%/bin-wrappers}" | tr -c "[a-zA-Z0-9]" "[_*]")"."
# make the tested dir absolute
GIT_TEST_INSTALLED=$(cd "$GIT_TEST_INSTALLED" && pwd)
fi
TEST_NO_CREATE_REPO=t
TEST_NO_MALLOC_CHECK=t
Introduce a performance testing framework This introduces a performance testing framework under t/perf/. It tries to be as close to the test-lib.sh infrastructure as possible, and thus should be easy to get used to for git developers. The following points were considered for the implementation: 1. You usually want to compare arbitrary revisions/build trees against each other. They may not have the performance test under consideration, or even the perf-lib.sh infrastructure. To cope with this, the 'run' script lets you specify arbitrary build dirs and revisions. It even automatically builds the revisions if it doesn't have them at hand yet. 2. Usually you would not want to run all tests. It would take too long anyway. The 'run' script lets you specify which tests to run; or you can also do it manually. There is a Makefile for discoverability and 'make clean', but it is not meant for real-world use. 3. Creating test repos from scratch in every test is extremely time-consuming, and shipping or downloading such large/weird repos is out of the question. We leave this decision to the user. Two different sizes of test repos can be configured, and the scripts just copy one or more of those (using hardlinks for the object store). By default it tries to use the build tree's git.git repository. This is fairly fast and versatile. Using a copy instead of a clone preserves many properties that the user may want to test for, such as lots of loose objects, unpacked refs, etc. Signed-off-by: Thomas Rast <trast@student.ethz.ch> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-02-17 18:25:09 +08:00
. ../test-lib.sh
# Variables from test-lib that are normally internal to the tests; we
# need to export them for test_perf subshells
export TEST_DIRECTORY TRASH_DIRECTORY GIT_BUILD_DIR GIT_TEST_CMP
Introduce a performance testing framework This introduces a performance testing framework under t/perf/. It tries to be as close to the test-lib.sh infrastructure as possible, and thus should be easy to get used to for git developers. The following points were considered for the implementation: 1. You usually want to compare arbitrary revisions/build trees against each other. They may not have the performance test under consideration, or even the perf-lib.sh infrastructure. To cope with this, the 'run' script lets you specify arbitrary build dirs and revisions. It even automatically builds the revisions if it doesn't have them at hand yet. 2. Usually you would not want to run all tests. It would take too long anyway. The 'run' script lets you specify which tests to run; or you can also do it manually. There is a Makefile for discoverability and 'make clean', but it is not meant for real-world use. 3. Creating test repos from scratch in every test is extremely time-consuming, and shipping or downloading such large/weird repos is out of the question. We leave this decision to the user. Two different sizes of test repos can be configured, and the scripts just copy one or more of those (using hardlinks for the object store). By default it tries to use the build tree's git.git repository. This is fairly fast and versatile. Using a copy instead of a clone preserves many properties that the user may want to test for, such as lots of loose objects, unpacked refs, etc. Signed-off-by: Thomas Rast <trast@student.ethz.ch> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-02-17 18:25:09 +08:00
perf_results_dir=$TEST_OUTPUT_DIRECTORY/test-results
mkdir -p "$perf_results_dir"
rm -f "$perf_results_dir"/$(basename "$0" .sh).subtests
if test -z "$GIT_PERF_REPEAT_COUNT"; then
GIT_PERF_REPEAT_COUNT=3
fi
die_if_build_dir_not_repo () {
if ! ( cd "$TEST_DIRECTORY/.." &&
git rev-parse --build-dir >/dev/null 2>&1 ); then
error "No $1 defined, and your build directory is not a repo"
fi
}
if test -z "$GIT_PERF_REPO"; then
die_if_build_dir_not_repo '$GIT_PERF_REPO'
GIT_PERF_REPO=$TEST_DIRECTORY/..
fi
if test -z "$GIT_PERF_LARGE_REPO"; then
die_if_build_dir_not_repo '$GIT_PERF_LARGE_REPO'
GIT_PERF_LARGE_REPO=$TEST_DIRECTORY/..
fi
test_perf_create_repo_from () {
test "$#" = 2 ||
error "bug in the test script: not 2 parameters to test-create-repo"
repo="$1"
source="$2"
source_git=$source/$(cd "$source" && git rev-parse --git-dir)
mkdir -p "$repo/.git"
(
cd "$repo/.git" &&
{ cp -Rl "$source_git/objects" . 2>/dev/null ||
cp -R "$source_git/objects" .; } &&
for stuff in "$source_git"/*; do
case "$stuff" in
*/objects|*/hooks|*/config)
;;
*)
cp -R "$stuff" . || break
;;
esac
done &&
cd .. &&
git init -q &&
mv .git/hooks .git/hooks-disabled 2>/dev/null
) || error "failed to copy repository '$source' to '$repo'"
}
# call at least one of these to establish an appropriately-sized repository
test_perf_default_repo () {
test_perf_create_repo_from "${1:-$TRASH_DIRECTORY}" "$GIT_PERF_REPO"
}
test_perf_large_repo () {
if test "$GIT_PERF_LARGE_REPO" = "$GIT_BUILD_DIR"; then
echo "warning: \$GIT_PERF_LARGE_REPO is \$GIT_BUILD_DIR." >&2
echo "warning: This will work, but may not be a sufficiently large repo" >&2
echo "warning: for representative measurements." >&2
fi
test_perf_create_repo_from "${1:-$TRASH_DIRECTORY}" "$GIT_PERF_LARGE_REPO"
}
test_checkout_worktree () {
git checkout-index -u -a ||
error "git checkout-index failed"
}
# Performance tests should never fail. If they do, stop immediately
immediate=t
test_run_perf_ () {
test_cleanup=:
test_export_="test_cleanup"
export test_cleanup test_export_
/usr/bin/time -f "%E %U %S" -o test_time.$i "$SHELL" -c '
. '"$TEST_DIRECTORY"/test-lib-functions.sh'
Introduce a performance testing framework This introduces a performance testing framework under t/perf/. It tries to be as close to the test-lib.sh infrastructure as possible, and thus should be easy to get used to for git developers. The following points were considered for the implementation: 1. You usually want to compare arbitrary revisions/build trees against each other. They may not have the performance test under consideration, or even the perf-lib.sh infrastructure. To cope with this, the 'run' script lets you specify arbitrary build dirs and revisions. It even automatically builds the revisions if it doesn't have them at hand yet. 2. Usually you would not want to run all tests. It would take too long anyway. The 'run' script lets you specify which tests to run; or you can also do it manually. There is a Makefile for discoverability and 'make clean', but it is not meant for real-world use. 3. Creating test repos from scratch in every test is extremely time-consuming, and shipping or downloading such large/weird repos is out of the question. We leave this decision to the user. Two different sizes of test repos can be configured, and the scripts just copy one or more of those (using hardlinks for the object store). By default it tries to use the build tree's git.git repository. This is fairly fast and versatile. Using a copy instead of a clone preserves many properties that the user may want to test for, such as lots of loose objects, unpacked refs, etc. Signed-off-by: Thomas Rast <trast@student.ethz.ch> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-02-17 18:25:09 +08:00
test_export () {
[ $# != 0 ] || return 0
test_export_="$test_export_\\|$1"
shift
test_export "$@"
}
'"$1"'
ret=$?
set | sed -n "s'"/'/'\\\\''/g"';s/^\\($test_export_\\)/export '"'&'"'/p" >test_vars
exit $ret' >&3 2>&4
eval_ret=$?
if test $eval_ret = 0 || test -n "$expecting_failure"
then
test_eval_ "$test_cleanup"
. ./test_vars || error "failed to load updated environment"
fi
if test "$verbose" = "t" && test -n "$HARNESS_ACTIVE"; then
echo ""
fi
return "$eval_ret"
}
test_perf () {
test "$#" = 3 && { test_prereq=$1; shift; } || test_prereq=
test "$#" = 2 ||
error "bug in the test script: not 2 or 3 parameters to test-expect-success"
export test_prereq
if ! test_skip "$@"
then
base=$(basename "$0" .sh)
echo "$test_count" >>"$perf_results_dir"/$base.subtests
echo "$1" >"$perf_results_dir"/$base.$test_count.descr
if test -z "$verbose"; then
echo -n "perf $test_count - $1:"
else
echo "perf $test_count - $1:"
fi
for i in $(test_seq 1 $GIT_PERF_REPEAT_COUNT); do
Introduce a performance testing framework This introduces a performance testing framework under t/perf/. It tries to be as close to the test-lib.sh infrastructure as possible, and thus should be easy to get used to for git developers. The following points were considered for the implementation: 1. You usually want to compare arbitrary revisions/build trees against each other. They may not have the performance test under consideration, or even the perf-lib.sh infrastructure. To cope with this, the 'run' script lets you specify arbitrary build dirs and revisions. It even automatically builds the revisions if it doesn't have them at hand yet. 2. Usually you would not want to run all tests. It would take too long anyway. The 'run' script lets you specify which tests to run; or you can also do it manually. There is a Makefile for discoverability and 'make clean', but it is not meant for real-world use. 3. Creating test repos from scratch in every test is extremely time-consuming, and shipping or downloading such large/weird repos is out of the question. We leave this decision to the user. Two different sizes of test repos can be configured, and the scripts just copy one or more of those (using hardlinks for the object store). By default it tries to use the build tree's git.git repository. This is fairly fast and versatile. Using a copy instead of a clone preserves many properties that the user may want to test for, such as lots of loose objects, unpacked refs, etc. Signed-off-by: Thomas Rast <trast@student.ethz.ch> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-02-17 18:25:09 +08:00
say >&3 "running: $2"
if test_run_perf_ "$2"
then
if test -z "$verbose"; then
echo -n " $i"
else
echo "* timing run $i/$GIT_PERF_REPEAT_COUNT:"
fi
else
test -z "$verbose" && echo
test_failure_ "$@"
break
fi
done
if test -z "$verbose"; then
echo " ok"
else
test_ok_ "$1"
fi
base="$perf_results_dir"/"$perf_results_prefix$(basename "$0" .sh)"."$test_count"
"$TEST_DIRECTORY"/perf/min_time.perl test_time.* >"$base".times
fi
echo >&3 ""
}
# We extend test_done to print timings at the end (./run disables this
# and does it after running everything)
test_at_end_hook_ () {
if test -z "$GIT_PERF_AGGREGATING_LATER"; then
( cd "$TEST_DIRECTORY"/perf && ./aggregate.perl $(basename "$0") )
fi
}
test_export () {
export "$@"
}