Re: Only 27% of reviewed-by tags are explicit, and much more

From: Felipe Contreras <felipe.contreras@gmail.com>
Date: Jun 19, 2021, 21:45 UTC
Message-ID: <60ce6584e36a3_d067c2084@natae.notmuch>
In-Reply-To: <CAPig+cRHBFodZ=ZJzpU208AWMSD-V6xH8zKZwmDPJk5eeu571w@mail.gmail.com>

Eric Sunshine wrote:

> On Fri, Jun 18, 2021 at 12:00 PM Felipe Contreras
> <felipe.contreras@gmail.com> wrote:

Show 20 quoted lines

> > I've updated the script to consider all responses to the cover letter
> > that start with 'Re: '.
> >
> > Anyway, with the updated script the explicit reviewed-bys are 40%, and
> > here are the stats:
> >
> >   5. Eric Sunshine: 38% (45/116)
> >
> > You got considerably more, from 17 to 45, but still pretty far from a
> > 100%.
> 
> The numbers produced by your script don't agree with my own
> investigative spelunking through my own mailbox. What I found is that,
> via 33 emails, I've given my Reviewed-by: to 133 patches[1]. If the
> 116 computed by your script is accurate, then that means that not all
> of my Reviewed-by:'s made it into the project, which is believable.
> Nevertheless, according to my own mailbox -- accounting for 133
> patches -- I have almost certainly given an explicit Reviewed-by: to
> all 116 of those patches your script found, which means the script
> output should probably be "100% (116/116)".

No, the fact that you gave your Reviewed-by doesn't mean Junio took the trailer, or even the patch.

Take for example this one:

https://lore.kernel.org/git/CAPig+cTgoD-GvpoBQ6tcGX4T2XhuKccJAZ40P76vxVD_PfEc9A@mail.gmail.com/

You said Reviewed-by, but Junio put Helped-by in the commit message of d228eea514 (worktree: accept -f as short for --force for removal, 2018-04-17).

Or this one:

https://lore.kernel.org/git/CAPig+cTcqSa6AfeMQivnSdL=y2+WWw2MtSavDciMc84RcKURMA@mail.gmail.com/

You gave it, but Junio didn't even mention you in any of the commits of this series, like 6b15595151 (t3200: unset core.logallrefupdates when testing reflog creation, 2018-06-22).

Same for:

https://lore.kernel.org/git/CAPig+cQ2NRO4yaFkcGmUpY3TZcWkdg-vu6d7Fq7JgHzYSkcRgg@mail.gmail.com/ https://lore.kernel.org/git/CAPig+cS4Bj4N8d1a29z8=f30owOec1pB=yF32ZUPmDH2Tu2kXA@mail.gmail.com/ https://lore.kernel.org/git/CAPig+cTaOTfwzodKSabdy9HFbF66RuEXwmvjZ8QuQVFMaVpA7w@mail.gmail.com/ https://lore.kernel.org/git/CAPig+cSC8RZJ-+uP=ZExVH2ZyexfQmLjzdjoBA7yuWkdYE4EGQ@mail.gmail.com/ https://lore.kernel.org/git/CAPig+cQ-yLnjrsB1E-7=UXfGzuJHat6YtfS8EVRNP2dcjj_6TA@mail.gmail.com/ https://lore.kernel.org/git/CAPig+cT0-ftZZyRORx-W2_Nit6XdgrpiyGS=pRjGtHoz1jW+Kg@mail.gmail.com/

These were never merged:

https://lore.kernel.org/git/20160216050915.GA5765@flurp.local/ https://lore.kernel.org/git/CAPig+cSSdGeMuV1XLqROXvSeYfmkNc_N7E_pzfJWdDR6wfD80A@mail.gmail.com/

This one Junio changed the title of the patch, so the script could not find it:

https://lore.kernel.org/git/CAPig+cSuCouNCuKa99mct4UMPykuMVy3+7sqB6y+v+UtP2oeTw@mail.gmail.com/

However, there were instances where my code did skip your mails because they were in a patch series that was sent as in-reply-to another patch series. My code did not consider those cases.

Taking a look at all the instances from all the people where a Reviewed-by is nested deep inside a thread I noticed that with very few exceptions almost all of them refer to the entire patch series. So I updated the code to simply consider all the replies in the thread.

I ran the numbers yet again and now it's 62% explicit Reviewed-bys.

You have 94% making you the second top explicit giver, there's still 4 commits that got your Reviewed-by implicitly, and 2 that my script could not find because the title was changed.

https://lore.kernel.org/git/CAPig+cQxAxFUFE8j2O7iaZoAby9ioNd6Wf1OVAr5qU7kTrQOyQ@mail.gmail.com/ https://lore.kernel.org/git/CAPig+cRgLyjNW+7EwXZB-=xNej=FR_1dqneR8VaaHzFaYHiOBA@mail.gmail.com/ https://lore.kernel.org/git/CAPig+cTFsZCowqNxmNtr1za+O6KjZmqJBZLGmUFd77rHmD+5pQ@mail.gmail.com/ https://lore.kernel.org/git/20150918222524.GA22410@flurp.local/

1. Jonathan Nieder: 89% (281/314)
2. Jeff King: 11% (28/248)
3. Stefan Beller: 69% (132/190)
4. Matthieu Moy: 67% (88/131)
5. Eric Sunshine: 94% (110/116)
6. Derrick Stolee: 71% (73/102)
7. Taylor Blau: 75% (63/83)
8. Michael Haggerty: 85% (47/55)
9. Elijah Newren: 89% (42/47)
10. Johannes Schindelin: 48% (17/35)
11. Jonathan Tan: 71% (23/32)
12. Nguyễn Thái Ngọc Duy: 33% (10/30)
13. Ronnie Sahlberg: 100% (16/16)
14. SZEDER Gábor: 0% (0/14)
15. Luke Diamand: 7% (1/13)
16. Felipe Contreras: 8% (1/12)
17. Johannes Sixt: 40% (4/10)
18. Ævar Arnfjörð Bjarmason: 66% (6/9)
19. Stefano Lattarini: 62% (5/8)
20. Torsten Bögershausen: 28% (2/7)

Although we have many people in the 60%, 70%, 80%, and 90% range, there's still plenty in 10% and 20%.

Here's the histogram:

https://i.imgur.com/jwqEp5H.png

Still pretty far from 100%.

Cheers.

-- 
Felipe Contreras