So I had an eyeball at the various graphs of activity on ohloh
(eg https://www.ohloh.net/p/parrot/commits/summary )
and thought that probably last month by the crude measure of "number of
commits", Perl 6 probably now had more activity than the Perl 5 core.
(ie just activity on git://perl5.git.perl.org/perl.git )
So I've written a number cruncher to check this, and it's surprising.
If you take MoarVM + NQP + Rakudo + Roast, and compare that with Perl 5,
there has been
more Perl 6 activity every month this year.
Not just September.
I've added a couple more columns for Parrot, and for the combined repository
for IronRuby & IronPython (git://github.com/IronLanguages/main)
These are figures for blead/nom/master:
MoarVM nqp rakudo roast (sum) perl 5 parrot Iron*
Sep 2013 276 116 168 139 699 528 7 48
Aug 2013 373 124 224 120 841 601 8 3
Jul 2013 166 137 254 95 652 592 5 0
Jun 2013 119 165 226 99 609 465 23 0
May 2013 98 147 220 100 565 528 19 1
Apr 2013 198 156 152 47 553 74 58 3
Mar 2013 111 74 185 64 434 228 45 5
Feb 2013 105 197 71 34 407 206 17 9
Jan 2013 135 101 145 89 470 282 123 0
Dec 2012 36 35 115 56 242 440 220 4
Nov 2012 25 22 115 71 233 612 25 0
Oct 2012 31 112 132 46 321 310 65 0
Sep 2012 11 91 105 28 235 400 184 2
Aug 2012 129 223 188 47 587 479 96 8
Jul 2012 257 190 293 124 864 288 44 11
Jun 2012 383 99 209 78 769 813 210 62
May 2012 99 161 256 148 664 426 366 27
Apr 2012 52 103 171 91 417 201 144 35
Mar 2012 0 23 175 100 298 242 243 36
Feb 2012 0 202 190 150 542 401 45 49
Jan 2012 0 138 202 369 709 520 117 30
These are figures for commits reachable from all currently mirrored branches:
MoarVM nqp rakudo roast (sum) perl 5 parrot Iron*
Sep 2013 329 143 173 139 784 629 7 51
Aug 2013 383 131 230 121 865 658 8 13
Jul 2013 166 139 262 102 669 640 14 10
Jun 2013 119 165 226 99 609 643 43 8
May 2013 98 147 222 100 567 747 19 5
Apr 2013 198 171 152 47 568 81 60 14
Mar 2013 111 124 219 64 518 287 52 5
Feb 2013 105 214 74 34 427 248 97 19
Jan 2013 135 102 148 90 475 359 227 4
Dec 2012 36 41 115 56 248 455 245 9
Nov 2012 25 22 115 71 233 690 52 2
Oct 2012 31 112 132 46 321 436 284 8
Sep 2012 11 92 108 28 239 434 219 3
Aug 2012 129 224 190 47 590 536 102 8
Jul 2012 257 192 297 124 870 321 54 14
Jun 2012 383 101 218 78 780 857 256 83
May 2012 99 165 260 148 672 605 531 38
Apr 2012 52 103 171 91 417 285 331 58
Mar 2012 0 46 179 100 325 278 337 63
Feb 2012 0 202 190 150 542 413 50 62
Jan 2012 0 138 213 369 720 542 181 30
Number cruncher attached.
I think that my figures differ from Ohloh's because I'm doing the calculation
"now", not each month, and git is attributing to months based on commit date.
So any work that was rebased and then merged will be attributed to the date
of rebasing. I think that this is partly why April is relatively barren in
Perl 5 land - because branch activity in April was rebased onto v5.19.0 in
May before merging.
Nicholas Clark
#!/usr/bin/perl -w
my $all = shift;
use POSIX qw(ctime mktime strftime);
sub count_revisions {
my ($repo, $after, $before, $all) = @_;
my @heads;
if ($all) {
@heads = `git --git-dir $repo show-ref --heads --hash`;
chomp @heads;
} else {
@heads = 'HEAD';
}
# Yes, I know I can do this with git rev-list but this one is easier to
# debug:
my @command = ('git', '--git-dir', $repo , 'log', '--format=oneline',
'--after', $after, '--before', $before, @heads);
# print "\n@command\n";
open my $fh, '-|', @command
or die "Can't start @command";
my $count = 0;
++$count while <$fh>;
close $fh
or die "Can't stop @command";
$count;
}
# This is a hack :-)
my $sum;
my @repos = (qw(MoarVM nqp rakudo roast), sub { $sum }, qw(perl parrot main));
my %map = (perl => 'perl 5', main => 'Iron*');
print " ";
foreach my $name (@repos) {
$name = '(sum)'
if ref $name;
printf "%8s", $map{$name} ? $map{$name} : $name;
}
print "\n";
my $now = time;
foreach my $year (2013, 2012) {
foreach my $mon (reverse 1..12) {
my $before_t = mktime(0, 0, 0, 1, $mon, $year - 1900);
# Is the end of the time range in the future?
next
if $before_t > $now;
my $before = ctime($before_t);
my $after = ctime(mktime(0, 0, 0, 1, $mon -1, $year - 1900));
$sum = 0;
print strftime("%b %Y", 0, 0, 0, 1, $mon - 1, $year - 1900);
foreach my $repo (qw(MoarVM nqp rakudo roast), sub { $sum },
qw(perl parrot main)) {
my $count;
if (ref $repo) {
$count = $repo->();
} else {
$count = count_revisions("$repo.git", $after, $before, $all);
$sum += $count;
}
printf "%8d", $count;
}
print "\n";
}
}
