[GH-ISSUE #4879] [Design] Create release/6.5.0 maintenance branch as backport base for #3728 #3030

New issue

Open

opened 2026-10-31 00:35:06 +00:00 by bsakaguchi · 1 comment

bsakaguchi commented

2026-10-31 00:35:06 +00:00

Owner

Originally created by @millerdev26-cmd on GitHub (Oct 17, 2026).
Original GitHub issue: https://github.com/morluto/rea/issues/4879

Decision needed

Please create a release/6.5.0 maintenance branch in this repository at tag rea-agents-6.5.0 (3b79af28e80785afce9f5998d8f03289402a1847), so a backport can be opened against it:

git push origin 3b79af28e80785afce9f5998d8f03289402a1847:refs/heads/release/6.5.0

Existing maintenance branches (release/6.0.0 … release/6.3.0) all point at their respective tags; release/6.5.0 is missing. A contributor branch cannot open a draft PR because GitHub requires the base branch to exist in morluto/rea.

Context and evidence

Upstream #3728 (3d2846f2371bc3e73b7069f713f6daa80f4db78c) is already in origin/main and in the 6.6.0 release-please branch. The 6.5.0 release predates it, so this is a release-line backport only.

On base rea-agents-6.5.0, the real Windows Ghidra lane fails because inspect_native_instruction.incoming_references includes the synthetic loader entry reference while xrefs excludes it:

AssertionError: Incoming references differ from xrefs source addresses or their order
actual:   [ 'Entry%20Point:0x0', '0x1400000a8', '0x1400000ac', '0x1400000194' ]
expected: [ '0x1400000a8', '0x1400000ac', '0x1400000194' ]

With the backport the same lane passes: npm run verify:ghidra:windows → ok:true, synthetic_entry_reference_excluded:true, incoming_xrefs_parity:true.
Prepared branch (fork of morluto/rea): https://github.com/millerdev26-cmd/rea/tree/release-6.5.0-fix3728 — three files, mirrors upstream #3728 (bridge/ghidra/ReaGhidraBridge.java, src/ghidra/GhidraAnalysisProfile.ts, src/ghidra/GhidraSnapshotCache.test.ts).
main is not a valid base for this branch: it already contains the fix, so the PR would be duplicated/conflicting.

Expected interaction with targets

Read existing files or captured data

Verification available so far

I have a prototype or observed results

Success criteria and open questions

Maintainer creates release/6.5.0 at the rea-agents-6.5.0 tag.
A draft PR is then opened with base release/6.5.0; the author readiness declarations are left for a human owner.

References

Upstream fix: https://github.com/morluto/rea/pull/3728 (commit 3d2846f2371bc3e73b7069f713f6daa80f4db78c)
Real Windows Ghidra lane: scripts/verify-real-ghidra-windows.mjs, assertion in scripts/lib/real-ghidra-compact-references.mjs

Originally created by @millerdev26-cmd on GitHub (Oct 17, 2026). Original GitHub issue: https://github.com/morluto/rea/issues/4879 ### Decision needed Please create a `release/6.5.0` maintenance branch in this repository at tag `rea-agents-6.5.0` (`3b79af28e80785afce9f5998d8f03289402a1847`), so a backport can be opened against it: ``` git push origin 3b79af28e80785afce9f5998d8f03289402a1847:refs/heads/release/6.5.0 ``` Existing maintenance branches (`release/6.0.0` … `release/6.3.0`) all point at their respective tags; `release/6.5.0` is missing. A contributor branch cannot open a draft PR because GitHub requires the base branch to exist in `morluto/rea`. ### Context and evidence - Upstream #3728 (`3d2846f2371bc3e73b7069f713f6daa80f4db78c`) is already in `origin/main` and in the 6.6.0 release-please branch. The 6.5.0 release predates it, so this is a release-line backport only. - On base `rea-agents-6.5.0`, the real Windows Ghidra lane fails because `inspect_native_instruction.incoming_references` includes the synthetic loader entry reference while `xrefs` excludes it: ``` AssertionError: Incoming references differ from xrefs source addresses or their order actual: [ 'Entry%20Point:0x0', '0x1400000a8', '0x1400000ac', '0x1400000194' ] expected: [ '0x1400000a8', '0x1400000ac', '0x1400000194' ] ``` - With the backport the same lane passes: `npm run verify:ghidra:windows` → `ok:true`, `synthetic_entry_reference_excluded:true`, `incoming_xrefs_parity:true`. - Prepared branch (fork of `morluto/rea`): https://github.com/millerdev26-cmd/rea/tree/release-6.5.0-fix3728 — three files, mirrors upstream #3728 (`bridge/ghidra/ReaGhidraBridge.java`, `src/ghidra/GhidraAnalysisProfile.ts`, `src/ghidra/GhidraSnapshotCache.test.ts`). - `main` is not a valid base for this branch: it already contains the fix, so the PR would be duplicated/conflicting. ### Expected interaction with targets - Read existing files or captured data ### Verification available so far - I have a prototype or observed results ### Success criteria and open questions - Maintainer creates `release/6.5.0` at the `rea-agents-6.5.0` tag. - A draft PR is then opened with base `release/6.5.0`; the author readiness declarations are left for a human owner. ### References - Upstream fix: https://github.com/morluto/rea/pull/3728 (commit `3d2846f2371bc3e73b7069f713f6daa80f4db78c`) - Real Windows Ghidra lane: `scripts/verify-real-ghidra-windows.mjs`, assertion in `scripts/lib/real-ghidra-compact-references.mjs`

bsakaguchi commented

2026-10-31 00:35:08 +00:00

Author

Owner

@XinwangTao commented on GitHub (Oct 18, 2026):

Makes sense — the release-line backport pattern is already established (release/6.0.0 … 6.3.0 all point at their tags), so 6.5.0 just needs the matching branch to keep the maintenance story consistent. The evidence is solid: real Windows Ghidra lane fails on the 6.5.0 base and passes with the three-file backport, and main is indeed not a valid base since it already contains the fix.

One suggestion: when creating the branch, it may be worth also documenting the backport policy (e.g. which lines are still maintained and the criteria for a fix being backported), so future contributors know whether release/6.5.0 is expected to receive more than just this one fix.

@XinwangTao commented on GitHub (Oct 18, 2026): Makes sense — the release-line backport pattern is already established (release/6.0.0 … 6.3.0 all point at their tags), so 6.5.0 just needs the matching branch to keep the maintenance story consistent. The evidence is solid: real Windows Ghidra lane fails on the 6.5.0 base and passes with the three-file backport, and main is indeed not a valid base since it already contains the fix. One suggestion: when creating the branch, it may be worth also documenting the backport policy (e.g. which lines are still maintained and the criteria for a fix being backported), so future contributors know whether release/6.5.0 is expected to receive more than just this one fix.

bsakaguchi referenced this issue

2026-10-31 00:40:20 +00:00

[GH-ISSUE #5654] [Tracker] Open PRs that can't merge yet: conflicts and red CI (October 2026) #3425

bsakaguchi referenced this issue

2026-10-31 01:07:38 +00:00

[PR #3030] Fix/android runtime fallback #5566