Contributors
Twenty-three people have merged something here. This page names the ones whose work you are about to depend on, and what specifically they built, because "thanks to all our contributors" tells you nothing about who to ask when the thing they wrote breaks at two in the morning.
Tomas Leitner
Wrote the first version in a weekend to stop a service falling over, then spent two years removing things from it. Reviews most pull requests. Owns the gate itself and the decision that it blocks rather than buffers.
Priya Ramanathan
Rewrote the shutdown path after finding the Wait-without-Close deadlock in production. Turned a hang into a named panic. Half the tests on this project exist because of that one week.
Wojciech Dabrowski
Brought the generics API across in v2 and did the work of keeping the v1 adapter honest. If your upgrade from v1 was uneventful, that was deliberate and it took four months.
Mei-Ling Chow
Owns CI. Added the race detector to every test rather than a nightly job, which is why the 2.1.0 regression was caught in a pull request instead of by a user.
Contributors with a merged change
23
Contributors with more than one
9
First-time contributors in the last year
6
Median time to first review
3 days
Pull requests merged from outside the core four
41
The Nine And The Fourteen
Nine people have merged more than one change. Fourteen merged exactly one and have not been back, and that is a normal and good outcome rather than a retention problem. Somebody hit a bug, fixed it, and left. The library is better and they got on with their work. A project that needs every contributor to become a regular is a project with a staffing plan instead of a scope.
"The deadlock was mine to fix because I was the one who found it at two in the morning. That is usually how ownership gets assigned around here.
"Priya Ramanathan
What a first contribution actually looks like
Open an issue before a pull request for anything touching the gate
Small fixes do not need an issue. Typos and doc bugs least of all
Tests run with the race detector locally or CI will fail them for you
One behaviour change per pull request. Two makes review twice as slow
Nobody will ask you to sign anything
Review Is The Bottleneck And We Say So
Median time to first review is three days. It has been as long as three weeks during two of the last four release cycles, both times because one person was doing every review. That is a known single point of failure, it is written in the maintainer notes, and the honest position is that it has not been solved yet rather than that it is fine.
