Repository navigation
tools/run_tests.sh: fix --select-class scope leak; make leak-sweep opt-in only - #44
Merged
Merged
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #44 +/- ##
=========================================
Coverage 38.01% 38.01%
Complexity 735 735
=========================================
Files 243 243
Lines 14863 14863
Branches 2449 2449
=========================================
Hits 5650 5650
+ Misses 8011 8008 -3
- Partials 1202 1205 +3 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
This was referenced Sep 5, 2026
Thrameos
added a commit
that referenced
this pull request
Sep 5, 2026
preliminary update racing ahead of onBeforeDownload CefDownloadItemTest.cancelDuringDownloadInvokesNCancel() flaked intermittently in CI across multiple unrelated backport PRs (#38, #39, #41, #42), always failing the same assertion: "onBeforeDownload was never invoked". Root-caused by reproducing it locally (100% deterministic once a low-risk, narrowly-scoped repro was possible -- see PR #44's run_tests.sh fix, needed first to run just this one test without also running the whole suite + leak-sweep) and adding temporary native fprintf instrumentation (not included in this commit) at DownloadHandler::OnBeforeDownload/ OnDownloadUpdated and ClientHandler::GetDownloadHandler. That instrumentation showed CEF can deliver a preliminary OnDownloadUpdated() notification -- for a valid, real CefDownloadItem -- *before* OnBeforeDownload() has been dispatched to the handler at all. This test's onDownloadUpdated() used to treat "not complete" on the very first update it saw as "must actively cancel it", immediately calling callback.cancel() and terminateTest() -- racing ahead of onBeforeDownload and ending the test before it ever got a chance to run, which is exactly what "onBeforeDownload was never invoked" captures. Fix: ignore any onDownloadUpdated() notification that arrives before this handler's own onBeforeDownload() has recorded its decision (gotBeforeDownload[0]). Verified via direct reproduction: failed 3/3 in isolation before the fix (including against a heavily-used local CEF profile where it failed 100% of the time), passed cleanly every time after (5+ consecutive solo runs, plus the full class together) with no change to downloadCompletesAndWritesRealFile's behavior. Co-Authored-By: Claude Sonnet 5 <[email protected]> Claude-Session: https://claude.ai/code/session_01HBFmQs6JDypYP5qdC99yDq
…t-in only --select-class was silently additive to the script's own hardcoded --select-package tests.junittests (JUnit ORs selectors rather than intersecting them), so a "run just this one class" invocation actually ran the whole 123-test suite -- discovered live while trying to reproduce the CefDownloadItemTest flake with a narrow, low-risk repro. Backports the fix already made for this on coverage/phase1-value-objects-phase2-handlers (commit 7678597), unmodified. Compounding that bug: the suite's leak-sweep/process-isolated tests (tagged "leak-sweep"/"process-isolated") spawn a pool of up to 16 concurrent isolated CEF subprocesses, each its own browser+renderer+GPU+zygote process tree. A "run just this one class" invocation that silently widened back out to the whole suite therefore also silently ran the full leak-sweep pass at the end -- fanning out to dozens of real windows/processes with no indication that was about to happen, and reliably exhausting this particular dev machine's memory (8G+swap) to the point of a full VM hang/crash, twice in one session. leak-sweep must only ever be triggered deliberately via tools/run_leak_sweep_isolated.sh; it's now excluded here unconditionally, with no flag on this script able to re-enable it. Co-Authored-By: Claude Sonnet 5 <[email protected]> Claude-Session: https://claude.ai/code/session_01HBFmQs6JDypYP5qdC99yDq
Thrameos
force-pushed
the
backport/run-tests-select-and-leaksweep-safety
branch
from
September 5, 2026 21:23
1084086 to
8ce45b9
Compare
2 of 3 tasks
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
1 participant
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.
Summary
--select-classwas silently additive to the script's hardcoded--select-package tests.junittests(JUnit ORs selectors rather than intersecting them), so a "run just this one class" invocation was actually running the whole 123-test suite -- discovered live while trying to reproduce theCefDownloadItemTestCI flake with a narrow, low-risk local repro. Backports the fix already made for this oncoverage/phase1-value-objects-phase2-handlers(commit7678597), unmodified.@Tag("leak-sweep")/@Tag("process-isolated")) spawn a pool of up to 16 concurrent isolated CEF subprocesses, each its own browser+renderer+GPU+zygote process tree. A "run just this one class" invocation that silently widened back out to the whole suite therefore also silently ran the full leak-sweep pass at the end -- fanning out to dozens of real windows/processes with no indication that was about to happen, reliably exhausting a dev machine's memory (8G+swap) to the point of a full VM hang/crash, twice in one session.leak-sweep/process-isolatedmust only ever be triggered deliberately viatools/run_leak_sweep_isolated.sh;run_tests.shnow excludes both tags unconditionally, with no flag on this script able to re-enable them.Test plan
bash -n tools/run_tests.sh(syntax check)--select-class tests.junittests.CefDownloadItemTestnow runs only that class's 2 tests (previously ran all 123 + leak-sweep)🤖 Generated with Claude Code
https://claude.ai/code/session_01HBFmQs6JDypYP5qdC99yDq