[webkit-dev] PSA: some WebKit project “pulse” activity metrics

sideshowbarker sideshowbarker at gmail.com
Sun May 26 20:23:41 PDT 2024


Some links with data and charts showing project commit/PR activity metrics:

- https://git-pulse.github.io/snapshots/?project=WebKit_WebKit
- https://git-pulse.github.io/snapshots/WebKit-WebKit-2024-05-26-pulse.html

>From those, some “good” metrics for the project that particularly stand out:

- 1123 PRs merged per month (3rd highest)
- 1224 commits per month (6th highest)
- 130 unique committers per month (9th highest)
- 9.42 commits per committer (on average) per month (9th highest)

Those are averages over the last 12 months. And the rankings are relative
to the 78 other extremely-high-activity projects tracked by this tool.

The only two projects that have more PRs merged per month are the Homebrew
project repos — but their PR numbers are so ridiculously higher than all other
projects that they’re essentially in a class by themselves.

https://git-pulse.github.io/snapshots/WebKit-WebKit-2024-05-26-pulse.html
is a report that has the specific per-month numbers (rather than just
averages) for the project — along with some charts of the data.

-----------------------------------------------------------------------

Caveat: I created the tool used for generating all this stuff, and you can
find the source at https://github.com/git-pulse/tools — and if you want to,
you can run it yourself locally in your WebKit repo, like this:

curl -fsSLO https://git-pulse.github.io/tools/git-pulse && bash ./git-pulse

(That generates output to the shell/terminal while it’s running, and then
generates HTML output when it’s done. And due to GitHub API throttling, it
takes a while — maybe 5 minutes or more — to run to completion.)

Caveat: In general the numbers this tool ends up generating seem accurate —
but I’m not a statistician and there may be some bugs in calculations I
have it using to generate its numbers.

-----------------------------------------------------------------------

And finally, some maybe-not-as-good metrics that stand out:

- open PRs: 946 (2nd highest)
- rate at which number of open PRs is increasing: 33 per month (#1 highest)

The only other project with more open PRs is the Web Platform Tests (WPT)
project — and statistically speaking, it’s almost the same: 957.

And beyond the fact that the rate at which the number of open PRs for the
WebKit project is highest among all 78 projects tracked by this tool — it’s
also increasing at nearly _twice_ the rate as the project with the
next-highest rate of increase (which is “only” 17 per month).

Among all the projects tracked, the _average_ rate of increase in open PRs
each month is zero — and almost 20 of the other projects actually have a
_decrease_ in the number of open PRs per month.

The WPT project — which has the highest number of open PRs — actually also
has the highest decrease in open PRs per month: It’s closing 15 more PRs per
month than are being opened each month — vs the WebKit project gaining 33 new
PRs per month (and over the last year, gaining 429 unclosed PRs in total).

I’m not stating those numbers as any kind of criticism or sign of anything
that necessarily needs fixing — I’m just noting how much those numbers
stand out relative to those of some other projects.

And again, see the caveat above: I’m not a statistician — and to the degree
that any of these numbers actually mean anything at all, I’m anyway
personally not qualified to offer any useful analysis on them from any kind
of actually-informed point of view.

-- 
https://sideshowbarker.net/


More information about the webkit-dev mailing list