Pr 1558

[JeremySun0731]

I added a focused unit test (ReflectionDiffBuilderAccessibilityTest) to demonstrate that
ReflectionDiffBuilder still forces reflective access to private fields even when
force-access is disabled.

This test shows that ReflectionDiffBuilder has not yet been migrated to the opt-in
accessibility controls introduced via AbstractReflection in PR #1558, and still relies
on legacy forced reflection behavior.

The test is intentionally minimal and serves as executable evidence of the current
behavior discussed in LANG-1711.

[garydgregory]

@JeremySun0731

This looks like it duplicates most of #1558

Instead you should create a PR against my branch, otherwise I can't tell what you've added or changed.

[JeremySun0731]

Thanks for the clarification!You're right — the PR was based on master, which makes the diff include most of #1558.
I'll rework this as a PR against your branch so that it only shows the additional test demonstrating the ReflectionDiffBuilder behavior.

[JeremySun0731]

I’ve split out the test into a separate commit and based it on
LANG-1711-reflection-diff-safe.
After cherry-picking, it turns out the test commit is already
contained in that branch, so GitHub shows no diff when comparing.
Functionally, the test-only change is identical and isolated as requested.
Please let me know if you’d prefer it to be submitted against a different base.

[garydgregory]

There are still 37 files changed and 16 commits which can't be right. You likely want the PR to be against my branch (in my fork).

[JeremySun0731]

Thanks for confirming — that makes sense.

I double-checked the history and confirmed that the test-only commit is already
present in your LANG-1711-reflection-diff-safe branch, which explains why GitHub
shows a large diff when comparing against apache:master and no diff when rebased.
Since the requested behavior is already covered upstream, I’m happy to close
this PR unless you’d prefer it to be submitted against a different base.

[garydgregory]

There are still 37 files changed and 16 commits which can't be right. You likely want the PR to be against my branch (in my fork).