The 'retry timer - proven peer reinsertion' test is flaky, I think because there's 0 margin built-in. I added an additional check to verify that the retry was triggered once after the first wait, and noticed that it rarely was, so I added some margin.
I'm not 100% sure that the changes I made make sense though, so do review.
This is an example output for a failing run of the old test:
# retry timer - proven peer reinsertion
not ok 1 - should be equal
---
actual: 1
expected: 2
operator: is
source: |
t.is(calls, 2)
----^
rt.destroy()
stack: |
./test/retry-timer.js:37:5
runNextTicks (node:internal/process/task_queues:60:5)
process.processTimers (node:internal/timers:511:9)
async Test._run (./node_modules/brittle/index.js:573:7)
` `
The 'retry timer - proven peer reinsertion' test is flaky, I think because there's 0 margin built-in. I added an additional check to verify that the retry was triggered once after the first wait, and noticed that it rarely was, so I added some margin.
I'm not 100% sure that the changes I made make sense though, so do review.
This is an example output for a failing run of the old test: