[bpamiri]

Summary

#4330 fixed withAdvisoryLock(transaction = true) so that it releases its MySQL/SQL Server lock when the callback ends the request with abort or redirectTo(). The transaction = true option is itself new in 4.2 (#4222), so no released version ever had that bug. A separate 4.2 "Fixed" entry would describe a problem nobody upgrading has seen.

• Removed changelog.d/4219-advisory-lock-abort-release.fixed.md.
• Added one sentence to changelog.d/tx-advisory-locks.added.md, the feature's own entry: "The lock is released even when the callback ends the request with abort or redirectTo()." This matches the locking.cfm docblock on develop b5ccb4e: the release runs in a finally that executes on those paths.

#4330's other fragment, 4219-testclient-get-timeout.added, stays as it is. tools/changelog-promote.sh --preview renders.

🤖 Generated with Claude Code

[bpamiri]

Review @ ae1f738: APPROVE (changelog only)

• Dropping 4219-advisory-lock-abort-release.fixed is right. withAdvisoryLock(transaction = true) arrived with feat(model): opt-in transaction-scoped advisory locks for withAdvisoryLock #4222 and isn't in any release: v4.1.2's locking.cfm has no transaction argument. So the abort-release bug fixed by fix(model): release withAdvisoryLock(transaction=true) lock when the callback aborts #4330 was never shipped, and a fixed entry would describe a bug users never had.
• The added sentence is accurate. "The lock is released even when the callback ends the request with abort or redirectTo()" matches develop's locking.cfm. The release sits in an inner finally with no transaction-action statement. Its docblock (:33) records that this runs on abort, redirectTo() and cflocation on Lucee, Adobe and BoxLang. The one exception, a request killed without running its finally, is an engine limit the docblock already notes.
• Nothing else in the fragment changes, and it keeps the convention of no trailing period after the link.

No findings.