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
The unhandledrejection event fires when a promise rejects without having a rejection handler attached yet. This indicates one of two scenarios:
We have some async bubble in between creation of the promise and awaiting it, where it can reject before we await. This is bad because it will (intentionally) crash node, but not awful because it can't cause a false positive.
There's a bug where the Promise is not awaited at all before the end of the test; this is a bigger problem as it can cause false positives.
We have a lint rule which ensures we don't drop promises on the floor, but it won't catch when we do something like const p = ... and then misuse it.
We should do something at runtime to at least try to catch these. I don't know what exactly, but maybe have it set a flag that's asserted to be false after the end of a "run". We'd probably want a short "sleep" between the end of the last test and the end of the run to give promises an opportunity to reject. There's no obvious place to put this right now, because different runtimes run things differently, but we should try to do something here.
The text was updated successfully, but these errors were encountered:
The unhandledrejection event fires when a promise rejects without having a rejection handler attached yet. This indicates one of two scenarios:
I suspect this happened in Uncaught exception in shader execution tests: #3157.Ben says this actually caused a false positive, so this was probably the other scenario.We have a lint rule which ensures we don't drop promises on the floor, but it won't catch when we do something like
const p = ...
and then misuse it.We should do something at runtime to at least try to catch these. I don't know what exactly, but maybe have it set a flag that's asserted to be false after the end of a "run". We'd probably want a short "sleep" between the end of the last test and the end of the run to give promises an opportunity to reject. There's no obvious place to put this right now, because different runtimes run things differently, but we should try to do something here.
The text was updated successfully, but these errors were encountered: