You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
sdstolworthy opened this issue
Sep 27, 2023
· 0 comments
Labels
tech-debtRefactors, unsavory workarounds or other technical decisions that should be revisited latertestingEngineering label for issues related to tests or test tooling
Our test runner uses taskkill to interrupt the Sandbox watcher. This kills the node process without allowing the event queue to drain, leaving unresolved promises. In Sandbox, this is manifested by the post-deployment sandbox events not finishing before the process is killed. This causes tests to fail which rely on the presence of the client configuration generated in the after deployment event handler.
Right now, this is naively addressed by hardcoding a wait time in the code. This should be refactored to allow a more flexible approach. Either our Process interactor should be able to send "wait" interactions, or we should find a more elegant way to kill the process on windows.
The text was updated successfully, but these errors were encountered:
sdstolworthy
added
tech-debt
Refactors, unsavory workarounds or other technical decisions that should be revisited later
testing
Engineering label for issues related to tests or test tooling
labels
Sep 27, 2023
tech-debtRefactors, unsavory workarounds or other technical decisions that should be revisited latertestingEngineering label for issues related to tests or test tooling
See #254 (comment)
and #254 (comment)
Our test runner uses
taskkill
to interrupt the Sandbox watcher. This kills the node process without allowing the event queue to drain, leaving unresolved promises. In Sandbox, this is manifested by the post-deployment sandbox events not finishing before the process is killed. This causes tests to fail which rely on the presence of the client configuration generated in the after deployment event handler.Right now, this is naively addressed by hardcoding a wait time in the code. This should be refactored to allow a more flexible approach. Either our Process interactor should be able to send "wait" interactions, or we should find a more elegant way to kill the process on windows.
The text was updated successfully, but these errors were encountered: