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
Currently result of validation other the linearization is represented as an error and passed to antithesis as Always assertion that this error is nil. However, in some cases some validations are not run at all, for example no watch validation will be run if we were not able to read the persisted requests.
From antithesis point of view this validation succeeded, so the uploaded artifacts for watch validation might include case that didn't do any watch requests. It would be better to introduce 3 state logic of: success, failure, unknown. Call assert always only for success/failure and don't call it in case unknown. This would be similar to what we already do for linearization, however there unknown means timeout which we also want to keep track off.
Why is this needed?
Have more accurate reports in antithesis that point to real successes/failures instead of pointing to unknown cases.
What would you like to be added?
Currently result of validation other the linearization is represented as an error and passed to antithesis as Always assertion that this error is nil. However, in some cases some validations are not run at all, for example no watch validation will be run if we were not able to read the persisted requests.
From antithesis point of view this validation succeeded, so the uploaded artifacts for watch validation might include case that didn't do any watch requests. It would be better to introduce 3 state logic of: success, failure, unknown. Call assert always only for success/failure and don't call it in case unknown. This would be similar to what we already do for linearization, however there unknown means timeout which we also want to keep track off.
Why is this needed?
Have more accurate reports in antithesis that point to real successes/failures instead of pointing to unknown cases.
/cc @henrybear327 @joshjms @nwnt @marcus-hodgson-antithesis
The text was updated successfully, but these errors were encountered: