Stop verification at the first unanswered request - #40
Merged
Merged
Conversation
A request left unanswered for 2000 ms now ends the run instead of being reported as one errored check, since a late reply can be matched to a later check by address and message type. The rule lives in the runner, which promotes any result carrying a TimeoutException into an abort, so a check that catches the timeout itself and returns an error result cannot absorb it either. The console names where the run stopped, the exit code stays 1, and a report requested with --report is still written and states that the verification is incomplete. The identity and protocol version reads at startup no longer swallow a timeout either. An error reply is still treated as information, so a device that answers R_VERSION with an error is still held to v1.
bruno-f-cruz
approved these changes
Sep 13, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A request left unanswered for 2000 ms now ends the run. Continuing is unsafe since the protocol carries no request identifier, so replies are correlated by address and message type alone, and a reply arriving after its deadline stays in the stream where it can satisfy a later check with a stale frame and a stale timestamp.
The rule lives in
Runner.RunAllAsync, the single point every result from every suite passes through. A timeout can otherwise arrive three ways: thrown out of a check, caught by a check's own handler and returned as anErrorResult, or returned directly by a helper such asRegisterHelpers.AssertReadableAsyncwhere no exception ever escapes. Placing the rule at the choke point covers all three and keeps test authors from having to know the concern exists.The identity and protocol version reads at startup no longer swallow a timeout either. An error reply is still information, so a device that answers
R_VERSIONwith an error is still held to v1.Verified on hardware
Found by banchan86 reviewing #38, where unplugging a device mid-run produced 93 timeout results in a report that otherwise looked complete. Reproduced here by unplugging a Behavior device mid-run, twice, with the same command.
The first row is what found the remaining hole. Two checks had caught the timeout themselves and recorded it as an error before a third finally aborted, which is why the rule inspects results rather than exceptions. The final run records 14 passed and 1 skipped, with nothing failed and nothing errored, so every check in the report ran against a device that was still answering.
Documentation
The article no longer says silence produces an error result, since silence now ends the run, and states what an unanswered request does and that the report is still written. The exit code section covers a run that cannot continue.