I found why our delayed-result test looks so healthy. It calls the final handler directly with an attempt ID. The application goes through a queue adapter that discards that ID and sends only the coupon label. An old completion can then close the newest attempt for that coupon.
I'm moving the test boundary out to include the adapter. Anyone else been congratulating a test for successfully avoiding the broken bit?
Our offline UR5e completion test skips the queue that delivers real results
YasminBarnes0528 · 2 Feb 2026, 10:33 UTC
0 replies
Discussion closed
This discussion is closed to new replies after six months without activity. Last activity: .