Currently, E2E tests are not being run at all in the CI, making it possible to unintentionally introduce bugs in scenarios that are already being covered in these tests.
Ideally the E2E tests would be run for each durabletask-go backend and for the Dapr Workflow engine. Already an issue has been found with sub-orchestration retry policies where they work correctly for Dapr Workflow but not for the durabletask-go backend.
An issue was recently discovered where retry policies for sub-orchestrations don't work because the durabletask-go/sqlite backend doesn't support instance ID reuse. Having an E2E test execution would have made that problem more apparent sooner.
Currently, E2E tests are not being run at all in the CI, making it possible to unintentionally introduce bugs in scenarios that are already being covered in these tests.
Ideally the E2E tests would be run for each
durabletask-gobackend and for the Dapr Workflow engine. Already an issue has been found with sub-orchestration retry policies where they work correctly for Dapr Workflow but not for thedurabletask-gobackend. An issue was recently discovered where retry policies for sub-orchestrations don't work because the durabletask-go/sqlite backend doesn't support instance ID reuse. Having an E2E test execution would have made that problem more apparent sooner.