> welcome views/opinions on methodology

This is an opportunity for me to (again) broadcast my strong opinion
that "maintainer" and "committer" should be upgrades to the "reviewer" role,
not to the "uploader" type of contributor.

The earliest evidence I found upon cursory search is this [1] one
(although the links no longer work).
I consider the "reviewer:" link to be more important than the "owner:" link 
there.

Looking at Gerrit search fields [2], I consider even "commentby:"
to be more important than "owner:" or "author:"
when it comes to evaluating who will be good (both in quality and throughput)
at deciding which change can be merged.

In short, Gerrit has more metadata than what is visible
in the git log your script is processing.

Vratko.

[1] https://lists.opendaylight.org/g/TSC/message/2679
[2] https://gerrit-review.googlesource.com/Documentation/user-search.html

-----Original Message-----
From: [email protected] <[email protected]> On Behalf Of Pim van Pelt via 
lists.fd.io
Sent: Tuesday, 14 April, 2026 20:34
To: [email protected]
Subject: [vpp-dev] Commit stats

Hoi folks,

In today's community meeting we had a conversation about committers, 
maintainers and 'active' versus 'emeritus'. I offered to take a look at 
the vpp repo and pull some stats. I asked Claude to help, and wrote a 
little analysis script in https://gerrit.fd.io/r/c/vpp/+/45534 which I'm 
not intending on checking in, but welcome views/opinions on methodology.

I scan all commits for --days time (10y currently), and look at all 
gerrits and commits as two different concepts. The script tries to be 
clever and correlate e-mail addresses and Signed-off-By strings into 
single Human names. It then looks at both the N-count of gerrits and 
merges, and the last activity date. I've applied a bit of heuristics to 
remove 'big sweeping refactors' from the stats, by touching more than 
200 files.

I define three cutoffs:
- committers who haven't merged in the last N days (730d)
- maintainers of a component with low change rate, but no activity might 
be OK for stable components (183d)
- maintainers who do not contribute to their component, which does have 
a higher change rate rate (365d)

Some findings:
1) committers are bimodal: Of the 15 total, 11 are active, and 4 are 
dormant at any cutoff (183, 265, 730, 1095 days)
2) 27 maintainers of components have not sent a gerrit in >1y
3) 19 maintainers own components to which they haven't committed in 
 >6mo, but others have

Some obligatory kudos:
A) Florin has made the most commits in 10 years: 2018 gerrits!
B) Damjan has made the most merges in 10 years: 3500 merges! (and #2 
commits at 1624 gerrits)

groet,
Pim

-- 
Pim van Pelt <[email protected]>
PBVP1-RIPE https://ipng.ch/

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#26967): https://lists.fd.io/g/vpp-dev/message/26967
Mute This Topic: https://lists.fd.io/mt/118828284/21656
Group Owner: [email protected]
Unsubscribe: https://lists.fd.io/g/vpp-dev/leave/14379924/21656/631435203/xyzzy 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

  • ... Pim van Pelt via lists.fd.io
    • ... Jim Thompson via lists.fd.io
    • ... Damjan Marion via lists.fd.io
      • ... Benoit Ganne (bganne) via lists.fd.io
        • ... Florin Coras via lists.fd.io
    • ... Vratko Polak -X (vrpolak - PANTHEON TECHNOLOGIES at Cisco) via lists.fd.io

Reply via email to