ReviewOS

also looking at this

zig-utils/zig-test-framework

Add retries, repeated runs, and flaky-test reporting

#25
Merged glennmichael123 wants to merge codex/issue-7-retries-repeat-flaky into main
15 files +719 -53
src/test_loader.zigmodified+63-17
Changes to src/test_loader.zig
@@ -26,6 +26,12 @@ pub const LoaderOptions = struct {
2626 junit_output: []const u8 = "test-results.xml",
2727 /// Global per-file timeout checked by the shared execution policy
2828 timeout_ms: ?u64 = null,
29 /// Additional process attempts after a failed test file.
30 retries: usize = 0,
31 /// Number of successful repetitions required for every selected file.
32 repeat: usize = 1,
33 /// Treat recovered failures as an unsuccessful run.
34 fail_on_flaky: bool = false,
2935 /// Optional output supplied by a CLI host. Embedded/test callers default
3036 /// to stderr so they do not interfere with Zig's stdout test protocol.
3137 reporter_writer: ?*std.Io.Writer = null,
@@ -85,6 +91,11 @@ pub fn runDiscoveredTests(
8591 };
8692 var results = reporter.TestResults.init(allocator);
8793 defer results.deinit();
94 var executed_cases: std.ArrayList(suite.TestCase) = .empty;
95 defer {
96 for (executed_cases.items) |*test_case| test_case.deinit(allocator);
97 executed_cases.deinit(allocator);
98 }
8899
89100 try events.emit(.{ .run_started = selected_files });
90101
@@ -120,23 +131,49 @@ pub fn runDiscoveredTests(
120131 try server.broadcast("test_start", json);
121132 }
122133
123 const start_time = compat.nanoTimestamp();
124 var result = try runTestFile(allocator, file.path, options);
125 const end_time = compat.nanoTimestamp();
126 const execution_time_ns: u64 = @intCast(end_time - start_time);
127 const timed_out = policy.timedOut(execution_time_ns);
128 if (timed_out) result = false;
129
130134 files_run += 1;
131135 var test_case = suite.TestCase.init(file.name, externalTestNoop);
132136 test_case.file = file.relative_path;
133 test_case.execution_time_ns = execution_time_ns;
134 test_case.status = if (result) .passed else .failed;
135 if (timed_out) {
136 test_case.error_message = "External test file exceeded the configured timeout";
137 } else if (!result) {
138 test_case.error_message = "External zig test process failed";
137 const repeat_count = @max(options.repeat, 1);
138 var attempt_number: usize = 0;
139 var had_failure = false;
140 var final_failure = false;
141
142 repetitions: for (1..repeat_count + 1) |repetition| {
143 for (1..options.retries + 2) |attempt_in_repetition| {
144 attempt_number += 1;
145 const start_time = compat.nanoTimestamp();
146 var passed = try runTestFile(allocator, file.path, options);
147 const duration_ns: u64 = @intCast(compat.nanoTimestamp() - start_time);
148 const timed_out = policy.timedOut(duration_ns);
149 if (timed_out) passed = false;
150 const error_message = if (timed_out)
151 try allocator.dupe(u8, "External test file exceeded the configured timeout")
152 else if (!passed)
153 try allocator.dupe(u8, "External zig test process failed")
154 else
155 null;
156 try test_case.attempts.append(allocator, .{
157 .number = attempt_number,
158 .repetition = repetition,
159 .status = if (passed) .passed else .failed,
160 .duration_ns = duration_ns,
161 .error_message = error_message,
162 });
163 test_case.execution_time_ns += duration_ns;
164
165 if (passed) break;
166 had_failure = true;
167 test_case.error_message = error_message;
168 if (attempt_in_repetition == options.retries + 1) {
169 final_failure = true;
170 break :repetitions;
171 }
172 }
139173 }
174 test_case.status = if (final_failure) .failed else if (had_failure) .flaky else .passed;
175 if (!final_failure) test_case.error_message = null;
176 try executed_cases.append(allocator, test_case);
140177 try results.addTest(&test_case);
141178 try events.emit(.{ .test_finished = &test_case });
142179 try events.emit(.{ .suite_finished = file.relative_path });
@@ -144,14 +181,20 @@ pub fn runDiscoveredTests(
144181 // Notify UI of test file end
145182 if (options.ui_server) |server| {
146183 var buffer: [512]u8 = undefined;
147 const status = if (result) "passed" else "failed";
148 const json = try std.fmt.bufPrint(&buffer, "", .{ file.name, status, execution_time_ns });
184 const json = try std.fmt.bufPrint(&buffer, "", .{
185 file.name,
186 @tagName(test_case.status),
187 test_case.execution_time_ns,
188 test_case.attempts.items.len,
189 });
149190 try server.broadcast("test_end", json);
150191 try server.broadcast("suite_end", try std.fmt.bufPrint(&buffer, "", .{file.name}));
151192 }
152193
153 if (result) {
194 if (test_case.status == .passed) {
154195 if (options.verbose) std.debug.print(" ✓ {s} passed\n", .{file.name});
196 } else if (test_case.status == .flaky) {
197 std.debug.print(" ~ {s} flaky after {d} attempts\n", .{ file.name, test_case.attempts.items.len });
155198 } else {
156199 std.debug.print(" ✗ {s} failed\n", .{file.name});
157200 if (policy.shouldStop(&results)) {
@@ -188,7 +231,7 @@ pub fn runDiscoveredTests(
188231 try events.emit(.{ .run_finished = &results });
189232 try reporters.flush();
190233
191 return results.failed == 0;
234 return results.failed == 0 and (!options.fail_on_flaky or results.flaky == 0);
192235}
193236
194237fn externalTestNoop(_: std.mem.Allocator) !void {}
@@ -292,6 +335,9 @@ test "LoaderOptions default values" {
292335 try std.testing.expectEqual(reporter.ReporterType.spec, options.reporter_type);
293336 try std.testing.expectEqualStrings("test-results.xml", options.junit_output);
294337 try std.testing.expectEqual(@as(?u64, null), options.timeout_ms);
338 try std.testing.expectEqual(@as(usize, 0), options.retries);
339 try std.testing.expectEqual(@as(usize, 1), options.repeat);
340 try std.testing.expect(!options.fail_on_flaky);
295341 try std.testing.expectEqual(@as(?*std.Io.Writer, null), options.reporter_writer);
296342 try std.testing.expectEqual(@as(?coverage.CoverageOptions, null), options.coverage_options);
297343}