From 1aa6fa19737c025dc7f1ca637cd63dd7475e74ad Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Mon, 20 Jul 2026 22:43:55 +0000
Subject: [PATCH 1/2] Add -XX:+EnableDynamicAgentLoading to test argLine
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Silences the JDK 21 'A Java agent has been loaded dynamically' warning
emitted when Lincheck self-attaches its instrumentation agent at runtime
(no static -javaagent is configured). That warning is written to the
native stderr, outside Surefire's wrapped System.err, and can
intermittently corrupt the encoded fork channel — surfacing as
'Corrupted channel by directly writing to native stream' followed by a
bogus 'There was a timeout in the fork' even though every test passed.

Mirrors the same guard added to the sibling BitcoinAddressFinder repo
(there triggered by Mockito's inline mock maker self-attaching byte-buddy).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y84TKMNfhbFGi7EoCEvP99
---
 llama/pom.xml | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/llama/pom.xml b/llama/pom.xml
index e1593628c..5d48c77ef 100644
--- a/llama/pom.xml
+++ b/llama/pom.xml
@@ -609,7 +609,19 @@ SPDX-License-Identifier: MIT
 				<groupId>org.apache.maven.plugins</groupId>
 				<artifactId>maven-surefire-plugin</artifactId>
 				<configuration>
-					<argLine>@{argLine} -Xmx2g -XX:ErrorFile=hs_err_pid%p.log -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=.</argLine>
+					<!--
+					  -XX:+EnableDynamicAgentLoading silences the JDK 21 "A Java agent has been
+					  loaded dynamically" warning emitted when Lincheck self-attaches its
+					  instrumentation agent at runtime (no static -javaagent is configured for it).
+					  That warning is written to the *native* stderr (outside Surefire's wrapped
+					  System.err) and can intermittently corrupt the encoded fork channel —
+					  surfacing as "Corrupted channel by directly writing to native stream" and then
+					  a bogus "There was a timeout in the fork", even though every test passed.
+					  Enabling the flag removes the native write so the channel stays clean. Mirrors
+					  the same guard in the sibling BitcoinAddressFinder repo (there triggered by
+					  Mockito's inline mock maker self-attaching byte-buddy).
+					-->
+					<argLine>@{argLine} -Xmx2g -XX:ErrorFile=hs_err_pid%p.log -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=. -XX:+EnableDynamicAgentLoading</argLine>
 					<!--
 					  Capture each test class's stdout/stderr into
 					  target/surefire-reports/<class>-output.txt. When a native crash

From 76218c98b5fbd66271ae22c04977fedab4950776 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Tue, 21 Jul 2026 08:51:59 +0000
Subject: [PATCH 2/2] Shorten EnableDynamicAgentLoading argLine comment to
 workspace policy pointer

Replace the inline rationale with a one-line comment pointing to the canonical
workspace policy (ci-test-diagnostics.md section 2.1) plus the root-cause
upstream bug (byte-buddy#1639), keeping the explanation in one synchronized place.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y84TKMNfhbFGi7EoCEvP99
---
 llama/pom.xml | 13 +------------
 1 file changed, 1 insertion(+), 12 deletions(-)

diff --git a/llama/pom.xml b/llama/pom.xml
index 5d48c77ef..ae43fecc1 100644
--- a/llama/pom.xml
+++ b/llama/pom.xml
@@ -609,18 +609,7 @@ SPDX-License-Identifier: MIT
 				<groupId>org.apache.maven.plugins</groupId>
 				<artifactId>maven-surefire-plugin</artifactId>
 				<configuration>
-					<!--
-					  -XX:+EnableDynamicAgentLoading silences the JDK 21 "A Java agent has been
-					  loaded dynamically" warning emitted when Lincheck self-attaches its
-					  instrumentation agent at runtime (no static -javaagent is configured for it).
-					  That warning is written to the *native* stderr (outside Surefire's wrapped
-					  System.err) and can intermittently corrupt the encoded fork channel —
-					  surfacing as "Corrupted channel by directly writing to native stream" and then
-					  a bogus "There was a timeout in the fork", even though every test passed.
-					  Enabling the flag removes the native write so the channel stays clean. Mirrors
-					  the same guard in the sibling BitcoinAddressFinder repo (there triggered by
-					  Mockito's inline mock maker self-attaching byte-buddy).
-					-->
+					<!-- -XX:+EnableDynamicAgentLoading: silences the JDK 21 byte-buddy self-attach agent warning that intermittently corrupts Surefire's fork channel ("Corrupted channel ..." / bogus "timeout in the fork"). See workspace policy ci-test-diagnostics.md section 2.1: https://github.com/bernardladenthin/workspace/blob/main/policies/ci-test-diagnostics.md - root cause: https://github.com/raphw/byte-buddy/issues/1639 -->
 					<argLine>@{argLine} -Xmx2g -XX:ErrorFile=hs_err_pid%p.log -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=. -XX:+EnableDynamicAgentLoading</argLine>
 					<!--
 					  Capture each test class's stdout/stderr into