Repository navigation
Flake batch 3: generous waits for merge-queue timing flakes - #1047
Merged
Merged
Conversation
lsp_diagnostics_test::move_file_reports_deleted_source_and_created_destination_for_config_file waited 2 s for the second of two asynchronous didChangeWatchedFiles notifications and failed on a loaded Linux runner with one received. The assertions on both events are unchanged. Co-authored-by: alfonso-aft <289616620+alfonso-aft@users.noreply.github.com>
…act test issue_826_trigram_footer_agrees_with_a_scan_answer expects the footer `trigram: loading (this answer read files directly, no index)`. Its search request starts reloading the saved trigram cache (SearchLaneStatus::recover_evicted_trigram -> trigger_search_index_reload_if_evicted) and waits at most 250 ms for it. On Linux and on an idle Windows runner the reload finishes and is installed inside the request (the index is resident and ready afterwards), so the footer relabels a ready-but-unused lane as `loading`. On a loaded CI runner the reload can outlast the wait; the receiver is still installed when the footer is written and it truthfully says `trigram: building`. That is a test setup race, not a product bug: a reload really is in flight. Add a hidden AppContext test hook that engages the existing query-reload cooldown, call it in the test, and assert the request left no index or receiver behind. The footer then always comes from the not-resident, valid-header observation, and the exact footer assertion is kept. Co-authored-by: alfonso-aft <289616620+alfonso-aft@users.noreply.github.com>
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Test-only fixes for merge-queue flakes, batched so they land in one queue run.
lsp_diagnostics_test::move_file_reports_deleted_source_and_created_destination_for_config_file: moving a config file sends two asynchronous didChangeWatchedFiles notifications; the test waited 2 s for the second and failed on a loaded Linux runner (PR Worker edits stop building a formatter sandbox on every call #1035's group) with one received. It now waits up to 30 s; both event assertions are unchanged.aft_search_contract_test::issue_826_trigram_footer_agrees_with_a_scan_answer(Windows, twice today): the search request itself started reloading the saved trigram cache; when that reload outlived the request's 250 ms wait on a loaded runner, the footer truthfully saidtrigram: buildinginstead of the expectedloading. The test now holds back query-triggered reloads with a test hook (it sets the existing reload cooldown) and asserts that the request started no load, so it always exercises the case it describes: a valid saved cache that is not resident. The exact footer assertion is unchanged; product behaviour is unchanged.