Ankur Mhatre
243cccd4ca
Every @Async behaviour that surprises people, asserted by a test and captured to docs/output/: the self-invocation trap, what CGLIB cannot override, the IllegalArgumentException a plain return type throws, the unbounded queue that makes max-size decoration, spring.task.execution.propagate-context (new in Boot 4.1.0), the two Executor beans that leave @Async on an unpooled SimpleAsyncTaskExecutor, and JEP 491 measured on JDK 21 against JDK 25.
25 lines
1.0 KiB
Bash
Executable File
25 lines
1.0 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
# Runs PinningProbe with the virtual-thread scheduler limited to two carrier threads, so that
|
|
# pinning (if it happened) would serialise the work into visible seconds.
|
|
#
|
|
# The probe is compiled on its own with --release 21 rather than as part of the module, so the
|
|
# same class file can be run on a JDK 21 and a JDK 25 JVM. That comparison is the whole point:
|
|
# JEP 491 landed in JDK 24, and the difference between the two runs is the feature.
|
|
#
|
|
# Usage: scripts/pinning-probe.sh <path-to-javac> <path-to-java> [more java binaries...]
|
|
set -euo pipefail
|
|
cd "$(dirname "$0")/.."
|
|
JAVAC_BIN="${1:-javac}"
|
|
shift || true
|
|
OUT=target/probe-classes
|
|
mkdir -p "$OUT"
|
|
"$JAVAC_BIN" --release 21 -d "$OUT" src/main/java/com/ankurm/async/PinningProbe.java
|
|
|
|
for JAVA_BIN in "${@:-java}"; do
|
|
echo "---------------------------------------------------------------"
|
|
"$JAVA_BIN" \
|
|
-Djdk.virtualThreadScheduler.parallelism=2 \
|
|
-Djdk.virtualThreadScheduler.maxPoolSize=2 \
|
|
-cp "$OUT" com.ankurm.async.PinningProbe
|
|
done
|