TKTNewProcessTaskRunnerTest>>testNewProcessTaskRunnerRunsInNewProcessEveryTime failed 1 time in the sweep sweep14-new14 of our VM ("Got 9 instead of 10"). It passed in the 5 other saved sweeps of 2026-11-17 (3 of our VM, 2 of the patched Cog). The source of the test says that it fails at random on the CI of Pharo. The cause is not proven.
"Run in a Playground of a Pharo 14 image (Do it and print each line)." ((TKTNewProcessTaskRunnerTest selector: #testNewProcessTaskRunnerRunsInNewProcessEveryTime) run) printString.
testNewProcessTaskRunnerRunsInNewProcessEveryTime | runner hashes futures hashesSet | self skipOnPharoCITestingEnvironment. "this fails randomly on the CI" runner := TKTNewProcessTaskRunner new. hashes := AtomicSharedQueue new. futures := (1 to: 10) collect: [ :i | (runner future: [ Processor activeProcess identityHash ]) onSuccessDo: [ :v | hashes nextPut: v ] ]. futures do: [ :each | each waitForCompletion: 2 seconds ]. hashesSet := Set new. [ hashes isEmpty ] whileFalse: [ hashesSet add: hashes next ]. self assert: hashesSet size equals: 10
None yet. A draft of a test change exists in the iospharo working tree. The draft is not validated, and its cause is a hypothesis, so it is not published here.
State of 2026-11-20. Made from the test records of the iospharo project (a second Pharo VM). The fixes are proposals for review.