Missing commits from the bbPress Git mirror

[netweb]

The bbPress Git mirror at is currently missing 655 commits

I've not been able to identify any specific commits that are missing that would aid discovering any pattern to reveal a possible cause of this issue.

The following table shows the number of commits per user missing from the Git repo compared to the SVN repo

Author	w.org username	SVN Root	Git	Missing Git commit count
John James Jacoby	johnjamesjacoby	3491	3208	-283
Michael Adams	mdawaffe	1162	1042	-120
Sam Bauers	sambauers	1085	893	-192
Stephen Edgar	netweb	576	557	-19
Matt Mullenweg	matt	150	149	-1
Chris Hajer	chrishajer	67	67	0
Alex Concha	xknown	26		-26
bumpbot	bumpbot	21	21	0
Ryan Boren	ryan	16	15	-1
Andrew Nacin	nacin	13	1	-12
Jennifer M. Dodd	jmdodd	12	12	0
Pete Mall	PeteMall	11	11	0
Potter_System	Potter_System	7	7	0
Sergey Biryukov	SergeyBiryukov	5	5	0
Austin Matzko	filosofo	3	3	0
Peter Westwood	westi	1		-1
Barry	barry	1	1	0
				-655
	Total	6647	5992	-655

The above calculations were calculated using the following methods:

SVN commits:

• Repo: https://bbpress.svn.wordpress.org
• Command: svn log -v --xml | grep '<author.*/author>' | sort | uniq -c | sort -rn (Ran from the root of the repo)
• Total number of commits reported: 6647

W.org Git repo commits:

• Repo: git://bbpress.git.wordpress.org/
• Command: git shortlog -s
• Total number of commits reported: 5992

GitHub Git repo commits:

• Repo: https://github.com/ntwb/bbpress-mirror-test
• Command: git shortlog -s
• Total number of commits reported: 5992

Notes:

• The GitHub repo at ​https://github.com/ntwb/bbpress-mirror-test is a mirror of git://bbpress.git.wordpress.org/ created by importing the repo using the GitHub ​https://github.com/new/import importer tool importing the git://bbpress.git.wordpress.org/ repo.

• The SVN repo at the time of creating this issue was at revision ​r6650

[dd32]

Went to take a look if this is still an issue.. it is, but it's not.

Some "null commits" such as ​https://bbpress.trac.wordpress.org/changeset/31 don't show up in git because well.. they can't.. there's nothing there.

But the original command here only covered commits to trunk rather than to branches.

Eg:

$ git remote -v
origin	git://bbpress.git.wordpress.org/ (fetch)
origin	git://bbpress.git.wordpress.org/ (push)

$ git shortlog -s | cut -f1 | paste -s -d+ - | bc
6409

$ git shortlog -s --all | cut -f1 | paste -s -d+ - | bc
7283

$ svn log https://bbpress.svn.wordpress.org --limit 1 | grep ^r
r7229 | johnjamesjacoby | 2021-11-29 15:28:59 +0000 (Mon, 29 Nov 2021) | 3 lines

So based on that, there's MORE commits to Git than to SVN.. which is probably due to commits to multiple branches in SVN being multiple commits in git?

Which is confirmed by:

$ git log --all 2>&1 | grep 'git-svn-id' | cut -d: -f2-3 | cut -d' ' -f2 | cut -d'@' -f2 | sort | uniq -c | sort  | tail -n10
      2 1790
      2 1900
      2 1901
      2 5446
      2 5611
      2 5970
      2 5990
      2 6825
      3 5698
      5 5690

​https://bbpress.trac.wordpress.org/changeset/5690 is referenced by 5 Git commits, as it touched 5 different git branches.

Closing as worksforme. Re-open if you can point to a missing commit.