perf: add a comparison test of log --grep regex engines with -F
authorÆvar Arnfjörð Bjarmason <avarab@gmail.com>
Thu, 25 May 2017 19:45:24 +0000 (19:45 +0000)
committerJunio C Hamano <gitster@pobox.com>
Fri, 26 May 2017 03:52:37 +0000 (12:52 +0900)
Add a performance comparison test of log --grepgrep regex engines
given fixed strings.

See the preceding fixed-string t/perf change ("perf: add a comparison
test of grep regex engines with -F", 2017-04-21) for notes about this,
in particular this mostly tests exactly the same codepath now, but
might not in the future:

$ GIT_PERF_REPEAT_COUNT=10 GIT_PERF_LARGE_REPO=~/g/linux ./run p4221-log-grep-engines-fixed.sh
[...]
Test this tree
--------------------------------------------------------
4221.1: fixed log --grep='int' 5.99(5.55+0.40)
4221.2: basic log --grep='int' 5.92(5.56+0.31)
4221.3: extended log --grep='int' 6.01(5.51+0.45)
4221.4: perl log --grep='int' 5.99(5.56+0.38)
4221.6: fixed log --grep='uncommon' 5.06(4.76+0.27)
4221.7: basic log --grep='uncommon' 5.02(4.78+0.21)
4221.8: extended log --grep='uncommon' 4.99(4.78+0.20)
4221.9: perl log --grep='uncommon' 5.00(4.72+0.26)
4221.11: fixed log --grep='æ' 5.35(5.12+0.20)
4221.12: basic log --grep='æ' 5.34(5.11+0.20)
4221.13: extended log --grep='æ' 5.39(5.10+0.22)
4221.14: perl log --grep='æ' 5.44(5.16+0.23)

Only the non-ASCII -i case is different:

$ GIT_PERF_REPEAT_COUNT=10 GIT_PERF_LARGE_REPO=~/g/linux GIT_PERF_4221_LOG_OPTS=' -i' ./run p4221-log-grep-engines-fixed.sh
[...]
Test this tree
-----------------------------------------------------------
4221.1: fixed log -i --grep='int' 6.17(5.77+0.35)
4221.2: basic log -i --grep='int' 6.16(5.59+0.39)
4221.3: extended log -i --grep='int' 6.15(5.70+0.39)
4221.4: perl log -i --grep='int' 6.15(5.69+0.38)
4221.6: fixed log -i --grep='uncommon' 5.10(4.88+0.21)
4221.7: basic log -i --grep='uncommon' 5.04(4.76+0.25)
4221.8: extended log -i --grep='uncommon' 5.07(4.82+0.23)
4221.9: perl log -i --grep='uncommon' 5.03(4.78+0.22)
4221.11: fixed log -i --grep='æ' 5.93(5.65+0.25)
4221.12: basic log -i --grep='æ' 5.88(5.62+0.25)
4221.13: extended log -i --grep='æ' 6.02(5.69+0.29)
4221.14: perl log -i --grep='æ' 5.36(5.06+0.29)

See commit ("perf: add a comparison test of grep regex engines",
2017-04-19) for details on the machine the above test run was executed
on.

Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
t/perf/p4221-log-grep-engines-fixed.sh [new file with mode: 0755]
diff --git a/t/perf/p4221-log-grep-engines-fixed.sh b/t/perf/p4221-log-grep-engines-fixed.sh
new file mode 100755 (executable)
index 0000000..0609712
--- /dev/null
@@ -0,0 +1,44 @@
+#!/bin/sh
+
+test_description="Comparison of git-log's --grep regex engines with -F
+
+Set GIT_PERF_4221_LOG_OPTS in the environment to pass options to
+git-grep. Make sure to include a leading space,
+e.g. GIT_PERF_4221_LOG_OPTS=' -i'. Some options to try:
+
+       -i
+       --invert-grep
+       -i --invert-grep
+"
+
+. ./perf-lib.sh
+
+test_perf_large_repo
+test_checkout_worktree
+
+for pattern in 'int' 'uncommon' 'æ'
+do
+       for engine in fixed basic extended perl
+       do
+               if test $engine = "perl" && ! test_have_prereq PCRE
+               then
+                       prereq="PCRE"
+               else
+                       prereq=""
+               fi
+               test_perf $prereq "$engine log$GIT_PERF_4221_LOG_OPTS --grep='$pattern'" "
+                       git -c grep.patternType=$engine log --pretty=format:%h$GIT_PERF_4221_LOG_OPTS --grep='$pattern' >'out.$engine' || :
+               "
+       done
+
+       test_expect_success "assert that all engines found the same for$GIT_PERF_4221_LOG_OPTS '$pattern'" '
+               test_cmp out.fixed out.basic &&
+               test_cmp out.fixed out.extended &&
+               if test_have_prereq PCRE
+               then
+                       test_cmp out.fixed out.perl
+               fi
+       '
+done
+
+test_done