[PATCH v3 0/7] Codespeed perf results

[PATCH v3 0/7] Codespeed perf results

[Christian Couder]

This patch series is built on top of cc/perf-run-config which recently
graduated to master.

It makes it possible to send perf results to a Codespeed server. See
https://github.com/tobami/codespeed/ and web sites like
http://speed.pypy.org/ which are using Codespeed.

The end goal would be to have such a server always available to track
how the different git commands perform over time on different kind of
repos (small, medium, large, ...) with different optimizations on and
off (split-index, libpcre2, BLK_SHA1, ...)

With this series and a config file like:

$ cat perf.conf
[perf]
        dirsOrRevs = v2.12.0 v2.13.0
        repeatCount = 10
        sendToCodespeed = http://localhost:8000
        repoName = Git repo
[perf "with libpcre"]
        makeOpts = "DEVELOPER=1 USE_LIBPCRE=YesPlease"
[perf "without libpcre"]
        makeOpts = "DEVELOPER=1"

One should be able to just launch:

$ ./run --config perf.conf p7810-grep.sh

and then get nice graphs in a Codespeed instance running on
http://localhost:8000.

Caveat
~~~~~~

For now one has to create the "Git repo" environment (in fact all the
values of the "environment" field sent to Codespeed) in the Codespeed
admin interface. (We send the perf.repoName config variable in the
"environment" Codespeed field.) This is because Codespeed requires the
environment fields to be created and does not provide a simple way to
create these fields programmatically.

There are discussions on a Codespeed issue
(https://github.com/tobami/codespeed/issues/232) about creating a
proper API for Codespeed that could address this problem in the
future.

Changes since previous version
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

There are very few changes compared to v2:

  - In patch 1/7 commit message has been improved following a comment
    by Junio.

  - In patch 3/7 some debugging comments were removed and 'use JSON;'
    was moved to the top of aggregate.perl as suggested by Junio.

The diff is the following:

diff --git a/t/perf/aggregate.perl b/t/perf/aggregate.perl
index 34d74fc015..5c439f6bc2 100755
--- a/t/perf/aggregate.perl
+++ b/t/perf/aggregate.perl
@@ -3,6 +3,7 @@
 use lib '../../perl/blib/lib';
 use strict;
 use warnings;
+use JSON;
 use Git;
 
 sub get_times {
@@ -226,10 +227,6 @@ sub print_codespeed_results {
                }
        }
 
-       #use Data::Dumper qw/ Dumper /;
-       #print Dumper(\@data);
-
-       use JSON;
        print to_json(\@data, {utf8 => 1, pretty => 1}), "\n";
 }

Links
~~~~~

This patch series:

https://github.com/chriscool/git/commits/codespeed

Previous versions:

v1: https://github.com/chriscool/git/commits/codespeed1
v2: https://github.com/chriscool/git/commits/codespeed10

Discussions:

v1: https://public-inbox.org/git/CAP8UFD3Q4h-aYBDABSPOW948LQYVydWZ1hLPAD+kr9ZpXVZiaQ@mail.gmail.com/
v2: https://public-inbox.org/git/20171226215908.425-1-chriscool@tuxfamily.org/

Discussions about the cc/perf-run-config patch series:

v1: https://public-inbox.org/git/20170713065050.19215-1-chriscool@tuxfamily.org/
v2: https://public-inbox.org/git/CAP8UFD2j-UFh+9awz91gtZ-jusq7EUOExMgURO59vpf29jXS4A@mail.gmail.com/


Christian Couder (7):
  perf/aggregate: fix checking ENV{GIT_PERF_SUBSECTION}
  perf/aggregate: refactor printing results
  perf/aggregate: implement codespeed JSON output
  perf/run: add conf_opts argument to get_var_from_env_or_config()
  perf/run: learn about perf.codespeedOutput
  perf/run: learn to send output to codespeed server
  perf/run: read GIT_PERF_REPO_NAME from perf.repoName

 t/perf/aggregate.perl | 160 +++++++++++++++++++++++++++++++++++---------------
 t/perf/run            |  31 ++++++++--
 2 files changed, 137 insertions(+), 54 deletions(-)

-- 
2.16.0.rc0.40.gbe5e688583

[PATCH v3 1/7] perf/aggregate: fix checking ENV{GIT_PERF_SUBSECTION}

[Christian Couder]

The way we check ENV{GIT_PERF_SUBSECTION} could trigger
comparison between undef and "" that may be flagged by
use of strict & warnings. Let's fix that.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
---
 t/perf/aggregate.perl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/t/perf/aggregate.perl b/t/perf/aggregate.perl
index e401208488..769d418708 100755
--- a/t/perf/aggregate.perl
+++ b/t/perf/aggregate.perl
@@ -70,7 +70,7 @@ if (not @tests) {
 }
 
 my $resultsdir = "test-results";
-if ($ENV{GIT_PERF_SUBSECTION} ne "") {
+if (exists $ENV{GIT_PERF_SUBSECTION} and $ENV{GIT_PERF_SUBSECTION} ne "") {
 	$resultsdir .= "/" . $ENV{GIT_PERF_SUBSECTION};
 }
 
-- 
2.16.0.rc0.40.gbe5e688583

[PATCH v3 2/7] perf/aggregate: refactor printing results

[Christian Couder]

As we want to implement another kind of output than
the current output for the perf test results, let's
refactor the existing code that outputs the results
in its own print_default_results() function.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
---
 t/perf/aggregate.perl | 96 +++++++++++++++++++++++++++------------------------
 1 file changed, 50 insertions(+), 46 deletions(-)

diff --git a/t/perf/aggregate.perl b/t/perf/aggregate.perl
index 769d418708..3609cb5dc3 100755
--- a/t/perf/aggregate.perl
+++ b/t/perf/aggregate.perl
@@ -100,13 +100,6 @@ sub read_descr {
 	return $line;
 }
 
-my %descrs;
-my $descrlen = 4; # "Test"
-for my $t (@subtests) {
-	$descrs{$t} = $shorttests{$t}.": ".read_descr("$resultsdir/$t.descr");
-	$descrlen = length $descrs{$t} if length $descrs{$t}>$descrlen;
-}
-
 sub have_duplicate {
 	my %seen;
 	for (@_) {
@@ -122,54 +115,65 @@ sub have_slash {
 	return 0;
 }
 
-my %newdirabbrevs = %dirabbrevs;
-while (!have_duplicate(values %newdirabbrevs)) {
-	%dirabbrevs = %newdirabbrevs;
-	last if !have_slash(values %dirabbrevs);
-	%newdirabbrevs = %dirabbrevs;
-	for (values %newdirabbrevs) {
-		s{^[^/]*/}{};
+sub print_default_results {
+	my %descrs;
+	my $descrlen = 4; # "Test"
+	for my $t (@subtests) {
+		$descrs{$t} = $shorttests{$t}.": ".read_descr("$resultsdir/$t.descr");
+		$descrlen = length $descrs{$t} if length $descrs{$t}>$descrlen;
 	}
-}
 
-my %times;
-my @colwidth = ((0)x@dirs);
-for my $i (0..$#dirs) {
-	my $d = $dirs[$i];
-	my $w = length (exists $dirabbrevs{$d} ? $dirabbrevs{$d} : $dirnames{$d});
-	$colwidth[$i] = $w if $w > $colwidth[$i];
-}
-for my $t (@subtests) {
-	my $firstr;
+	my %newdirabbrevs = %dirabbrevs;
+	while (!have_duplicate(values %newdirabbrevs)) {
+		%dirabbrevs = %newdirabbrevs;
+		last if !have_slash(values %dirabbrevs);
+		%newdirabbrevs = %dirabbrevs;
+		for (values %newdirabbrevs) {
+			s{^[^/]*/}{};
+		}
+	}
+
+	my %times;
+	my @colwidth = ((0)x@dirs);
 	for my $i (0..$#dirs) {
 		my $d = $dirs[$i];
-		$times{$prefixes{$d}.$t} = [get_times("$resultsdir/$prefixes{$d}$t.times")];
-		my ($r,$u,$s) = @{$times{$prefixes{$d}.$t}};
-		my $w = length format_times($r,$u,$s,$firstr);
+		my $w = length (exists $dirabbrevs{$d} ? $dirabbrevs{$d} : $dirnames{$d});
 		$colwidth[$i] = $w if $w > $colwidth[$i];
-		$firstr = $r unless defined $firstr;
 	}
-}
-my $totalwidth = 3*@dirs+$descrlen;
-$totalwidth += $_ for (@colwidth);
-
-binmode STDOUT, ":utf8" or die "PANIC on binmode: $!";
+	for my $t (@subtests) {
+		my $firstr;
+		for my $i (0..$#dirs) {
+			my $d = $dirs[$i];
+			$times{$prefixes{$d}.$t} = [get_times("$resultsdir/$prefixes{$d}$t.times")];
+			my ($r,$u,$s) = @{$times{$prefixes{$d}.$t}};
+			my $w = length format_times($r,$u,$s,$firstr);
+			$colwidth[$i] = $w if $w > $colwidth[$i];
+			$firstr = $r unless defined $firstr;
+		}
+	}
+	my $totalwidth = 3*@dirs+$descrlen;
+	$totalwidth += $_ for (@colwidth);
 
-printf "%-${descrlen}s", "Test";
-for my $i (0..$#dirs) {
-	my $d = $dirs[$i];
-	printf "   %-$colwidth[$i]s", (exists $dirabbrevs{$d} ? $dirabbrevs{$d} : $dirnames{$d});
-}
-print "\n";
-print "-"x$totalwidth, "\n";
-for my $t (@subtests) {
-	printf "%-${descrlen}s", $descrs{$t};
-	my $firstr;
+	printf "%-${descrlen}s", "Test";
 	for my $i (0..$#dirs) {
 		my $d = $dirs[$i];
-		my ($r,$u,$s) = @{$times{$prefixes{$d}.$t}};
-		printf "   %-$colwidth[$i]s", format_times($r,$u,$s,$firstr);
-		$firstr = $r unless defined $firstr;
+		printf "   %-$colwidth[$i]s", (exists $dirabbrevs{$d} ? $dirabbrevs{$d} : $dirnames{$d});
 	}
 	print "\n";
+	print "-"x$totalwidth, "\n";
+	for my $t (@subtests) {
+		printf "%-${descrlen}s", $descrs{$t};
+		my $firstr;
+		for my $i (0..$#dirs) {
+			my $d = $dirs[$i];
+			my ($r,$u,$s) = @{$times{$prefixes{$d}.$t}};
+			printf "   %-$colwidth[$i]s", format_times($r,$u,$s,$firstr);
+			$firstr = $r unless defined $firstr;
+		}
+		print "\n";
+	}
 }
+
+binmode STDOUT, ":utf8" or die "PANIC on binmode: $!";
+
+print_default_results();
-- 
2.16.0.rc0.40.gbe5e688583

[PATCH v3 3/7] perf/aggregate: implement codespeed JSON output

[Christian Couder]

Codespeed (https://github.com/tobami/codespeed/) is an open source
project that can be used to track how some software performs over
time. It stores performance test results in a database and can show
nice graphs and charts on a web interface.

As it can be interesting to use Codespeed to see how Git performance
evolves over time and releases, let's implement a Codespeed output
in "perf/aggregate.perl".

Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
---
 t/perf/aggregate.perl | 64 +++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 62 insertions(+), 2 deletions(-)

diff --git a/t/perf/aggregate.perl b/t/perf/aggregate.perl
index 3609cb5dc3..5c439f6bc2 100755
--- a/t/perf/aggregate.perl
+++ b/t/perf/aggregate.perl
@@ -3,6 +3,7 @@
 use lib '../../perl/blib/lib';
 use strict;
 use warnings;
+use JSON;
 use Git;
 
 sub get_times {
@@ -35,10 +36,15 @@ sub format_times {
 	return $out;
 }
 
-my (@dirs, %dirnames, %dirabbrevs, %prefixes, @tests);
+my (@dirs, %dirnames, %dirabbrevs, %prefixes, @tests, $codespeed);
 while (scalar @ARGV) {
 	my $arg = $ARGV[0];
 	my $dir;
+	if ($arg eq "--codespeed") {
+		$codespeed = 1;
+		shift @ARGV;
+		next;
+	}
 	last if -f $arg or $arg eq "--";
 	if (! -d $arg) {
 		my $rev = Git::command_oneline(qw(rev-parse --verify), $arg);
@@ -70,8 +76,10 @@ if (not @tests) {
 }
 
 my $resultsdir = "test-results";
+my $results_section = "";
 if (exists $ENV{GIT_PERF_SUBSECTION} and $ENV{GIT_PERF_SUBSECTION} ne "") {
 	$resultsdir .= "/" . $ENV{GIT_PERF_SUBSECTION};
+	$results_section = $ENV{GIT_PERF_SUBSECTION};
 }
 
 my @subtests;
@@ -174,6 +182,58 @@ sub print_default_results {
 	}
 }
 
+sub print_codespeed_results {
+	my ($results_section) = @_;
+
+	my $project = "Git";
+
+	my $executable = `uname -s -m`;
+	chomp $executable;
+
+	if ($results_section ne "") {
+		$executable .= ", " . $results_section;
+	}
+
+	my $environment;
+	if (exists $ENV{GIT_PERF_REPO_NAME} and $ENV{GIT_PERF_REPO_NAME} ne "") {
+		$environment = $ENV{GIT_PERF_REPO_NAME};
+	} elsif (exists $ENV{GIT_TEST_INSTALLED} and $ENV{GIT_TEST_INSTALLED} ne "") {
+		$environment = $ENV{GIT_TEST_INSTALLED};
+		$environment =~ s|/bin-wrappers$||;
+	} else {
+		$environment = `uname -r`;
+		chomp $environment;
+	}
+
+	my @data;
+
+	for my $t (@subtests) {
+		for my $d (@dirs) {
+			my $commitid = $prefixes{$d};
+			$commitid =~ s/^build_//;
+			$commitid =~ s/\.$//;
+			my ($result_value, $u, $s) = get_times("$resultsdir/$prefixes{$d}$t.times");
+
+			my %vals = (
+				"commitid" => $commitid,
+				"project" => $project,
+				"branch" => $dirnames{$d},
+				"executable" => $executable,
+				"benchmark" => $shorttests{$t} . " " . read_descr("$resultsdir/$t.descr"),
+				"environment" => $environment,
+				"result_value" => $result_value,
+			    );
+			push @data, \%vals;
+		}
+	}
+
+	print to_json(\@data, {utf8 => 1, pretty => 1}), "\n";
+}
+
 binmode STDOUT, ":utf8" or die "PANIC on binmode: $!";
 
-print_default_results();
+if ($codespeed) {
+	print_codespeed_results($results_section);
+} else {
+	print_default_results();
+}
-- 
2.16.0.rc0.40.gbe5e688583

[PATCH v3 4/7] perf/run: add conf_opts argument to get_var_from_env_or_config()

[Christian Couder]

Let's make it possible to use `git config` type specifiers like
`--int` or `--bool`, so that config values are converted to the
canonical form and easier to use.

This additional argument is now the fourth argument of
get_var_from_env_or_config() instead of the fifth because we
want the default value argument to be unset if it is not
passed, and this is simpler if it is the last argument.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
---
 t/perf/run | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/t/perf/run b/t/perf/run
index 43e4de49ef..214d658172 100755
--- a/t/perf/run
+++ b/t/perf/run
@@ -105,7 +105,8 @@ get_var_from_env_or_config () {
 	env_var="$1"
 	conf_sec="$2"
 	conf_var="$3"
-	# $4 can be set to a default value
+	conf_opts="$4" # optional
+	# $5 can be set to a default value
 
 	# Do nothing if the env variable is already set
 	eval "test -z \"\${$env_var+x}\"" || return
@@ -116,18 +117,18 @@ get_var_from_env_or_config () {
 	if test -n "$GIT_PERF_SUBSECTION"
 	then
 		var="$conf_sec.$GIT_PERF_SUBSECTION.$conf_var"
-		conf_value=$(git config -f "$GIT_PERF_CONFIG_FILE" "$var") &&
+		conf_value=$(git config $conf_opts -f "$GIT_PERF_CONFIG_FILE" "$var") &&
 		eval "$env_var=\"$conf_value\"" && return
 	fi
 	var="$conf_sec.$conf_var"
-	conf_value=$(git config -f "$GIT_PERF_CONFIG_FILE" "$var") &&
+	conf_value=$(git config $conf_opts -f "$GIT_PERF_CONFIG_FILE" "$var") &&
 	eval "$env_var=\"$conf_value\"" && return
 
-	test -n "${4+x}" && eval "$env_var=\"$4\""
+	test -n "${5+x}" && eval "$env_var=\"$5\""
 }
 
 run_subsection () {
-	get_var_from_env_or_config "GIT_PERF_REPEAT_COUNT" "perf" "repeatCount" 3
+	get_var_from_env_or_config "GIT_PERF_REPEAT_COUNT" "perf" "repeatCount" "--int" 3
 	export GIT_PERF_REPEAT_COUNT
 
 	get_var_from_env_or_config "GIT_PERF_DIRS_OR_REVS" "perf" "dirsOrRevs"
-- 
2.16.0.rc0.40.gbe5e688583

[PATCH v3 5/7] perf/run: learn about perf.codespeedOutput

[Christian Couder]

Let's make it possible to set in a config file the output
format (regular or codespeed) of the perf tests.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
---
 t/perf/run | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/t/perf/run b/t/perf/run
index 214d658172..4e62d6bb3f 100755
--- a/t/perf/run
+++ b/t/perf/run
@@ -144,10 +144,15 @@ run_subsection () {
 		set -- . "$@"
 	fi
 
+	codespeed_opt=
+	test "$GIT_PERF_CODESPEED_OUTPUT" = "true" && codespeed_opt="--codespeed"
+
 	run_dirs "$@"
-	./aggregate.perl "$@"
+	./aggregate.perl $codespeed_opt "$@"
 }
 
+get_var_from_env_or_config "GIT_PERF_CODESPEED_OUTPUT" "perf" "codespeedOutput" "--bool"
+
 cd "$(dirname $0)"
 . ../../GIT-BUILD-OPTIONS
 
-- 
2.16.0.rc0.40.gbe5e688583

[PATCH v3 6/7] perf/run: learn to send output to codespeed server

[Christian Couder]

Let's make it possible to set in a config file the URL of
a codespeed server. And then let's make the `run` script
send the perf test results to this URL at the end of the
tests.

This should make is possible to easily automate the process
of running perf tests and having their results available in
Codespeed.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
---
 t/perf/run | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/t/perf/run b/t/perf/run
index 4e62d6bb3f..ef56396546 100755
--- a/t/perf/run
+++ b/t/perf/run
@@ -148,10 +148,20 @@ run_subsection () {
 	test "$GIT_PERF_CODESPEED_OUTPUT" = "true" && codespeed_opt="--codespeed"
 
 	run_dirs "$@"
-	./aggregate.perl $codespeed_opt "$@"
+
+	if test -z "$GIT_PERF_SEND_TO_CODESPEED"
+	then
+		./aggregate.perl $codespeed_opt "$@"
+	else
+		json_res_file="test-results/$GIT_PERF_SUBSECTION/aggregate.json"
+		./aggregate.perl --codespeed "$@" | tee "$json_res_file"
+		send_data_url="$GIT_PERF_SEND_TO_CODESPEED/result/add/json/"
+		curl -v --request POST --data-urlencode "json=$(cat "$json_res_file")" "$send_data_url"
+	fi
 }
 
 get_var_from_env_or_config "GIT_PERF_CODESPEED_OUTPUT" "perf" "codespeedOutput" "--bool"
+get_var_from_env_or_config "GIT_PERF_SEND_TO_CODESPEED" "perf" "sendToCodespeed"
 
 cd "$(dirname $0)"
 . ../../GIT-BUILD-OPTIONS
-- 
2.16.0.rc0.40.gbe5e688583

[PATCH v3 7/7] perf/run: read GIT_PERF_REPO_NAME from perf.repoName

[Christian Couder]

The GIT_PERF_REPO_NAME env variable is used in
the `aggregate.perl` script to set the 'environment'
field in the JSON Codespeed output.

Let's make it easy to set this variable by setting it
in a config file.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
---
 t/perf/run | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/t/perf/run b/t/perf/run
index ef56396546..1a100d6134 100755
--- a/t/perf/run
+++ b/t/perf/run
@@ -137,6 +137,9 @@ run_subsection () {
 	get_var_from_env_or_config "GIT_PERF_MAKE_COMMAND" "perf" "makeCommand"
 	get_var_from_env_or_config "GIT_PERF_MAKE_OPTS" "perf" "makeOpts"
 
+	get_var_from_env_or_config "GIT_PERF_REPO_NAME" "perf" "repoName"
+	export GIT_PERF_REPO_NAME
+
 	GIT_PERF_AGGREGATING_LATER=t
 	export GIT_PERF_AGGREGATING_LATER
 
-- 
2.16.0.rc0.40.gbe5e688583