morluto/rea

Fork 0

mirror of https://github.com/morluto/rea.git synced 2026-11-05 07:47:29 +00:00

[GH-ISSUE #2138] [Feature] Complete the upstream session evidence in headless IDA cleanup diagnostics #698

New issue

Closed

opened 2026-10-15 18:27:29 +00:00 by bsakaguchi · 2 comments

bsakaguchi commented

2026-10-15 18:27:29 +00:00

Owner

Originally created by @AS9929 on GitHub (Oct 12, 2026).
Original GitHub issue: https://github.com/morluto/rea/issues/2138

Before submitting

I searched the open and closed issues and pull requests for a related proposal.
I removed credentials, tokens, private paths, and sensitive target data.

Area

Native analysis or decompilation (headless IDA provider)

Current limitation

#1841 adds an upstream_sessions entry to headless IDA cleanup_incomplete diagnostics. It names upstream sessions whose release REA could not confirm, each with session_id, backend and owned. To keep that PR focused on its core fix (one-shot CLI exit after retained cleanup), two refinements raised in review were deferred:

Startup inventory is not used as evidence. #initialize validates only the requested owned worker in the startup idb_list. If that inventory also lists another same-input session and the first idb_close then rejects, the diagnostic falls back to the requested worker with owned: null. The peer is omitted, and the ownership REA already parsed is dropped. (review thread)
Observed input paths are not reported. The post-close remaining filter keeps an entry whose session_id matches the requested database even when its input_path differs. The diagnostic drops that path, and its reason text presents the entry as a session on the private input. With mismatched or reused IDs, an operator cannot tell the unreleased worker from another database. (review thread)

Use cases

An operator or agent reads cleanup_incomplete after an interrupted headless IDA analysis to decide which upstream session to release, for example with idb_close, and which ones are external.

Desired outcome and success criteria

Session evidence from a successful startup inventory is retained as blocker evidence, until a later successful inventory replaces it.
Each observed session reports its input_path, or an explicit binding conflict when its ID matches but its path does not.
Tests cover a startup peer with a rejected first close, and a post-close entry with the requested ID but a different input path.

Authority boundaries and non-goals

Diagnostics only. Nothing here closes, kills or adopts upstream sessions, and successful-close validation is not weakened.

Alternatives considered

Folding these into #1841. They were deferred because each review round in that PR raised further diagnostic edge cases, while its core fix is the CLI exit.

Possible direction

Record the startup idb_list same-input entries through the same blocker-recording path used for recovery and post-close inventories, and carry input_path in UpstreamSession.

References

#1841, #1710

Originally created by @AS9929 on GitHub (Oct 12, 2026). Original GitHub issue: https://github.com/morluto/rea/issues/2138 ### Before submitting - [x] I searched the open and closed issues and pull requests for a related proposal. - [x] I removed credentials, tokens, private paths, and sensitive target data. ### Area Native analysis or decompilation (headless IDA provider) ### Current limitation #1841 adds an `upstream_sessions` entry to headless IDA `cleanup_incomplete` diagnostics. It names upstream sessions whose release REA could not confirm, each with `session_id`, `backend` and `owned`. To keep that PR focused on its core fix (one-shot CLI exit after retained cleanup), two refinements raised in review were deferred: 1. **Startup inventory is not used as evidence.** `#initialize` validates only the requested owned worker in the startup `idb_list`. If that inventory also lists another same-input session and the first `idb_close` then rejects, the diagnostic falls back to the requested worker with `owned: null`. The peer is omitted, and the ownership REA already parsed is dropped. ([review thread](https://github.com/morluto/rea/pull/1841#discussion_r4246550413)) 2. **Observed input paths are not reported.** The post-close `remaining` filter keeps an entry whose `session_id` matches the requested database even when its `input_path` differs. The diagnostic drops that path, and its reason text presents the entry as a session on the private input. With mismatched or reused IDs, an operator cannot tell the unreleased worker from another database. ([review thread](https://github.com/morluto/rea/pull/1841#discussion_r4246550738)) ### Use cases An operator or agent reads `cleanup_incomplete` after an interrupted headless IDA analysis to decide which upstream session to release, for example with `idb_close`, and which ones are external. ### Desired outcome and success criteria - Session evidence from a successful startup inventory is retained as blocker evidence, until a later successful inventory replaces it. - Each observed session reports its `input_path`, or an explicit binding conflict when its ID matches but its path does not. - Tests cover a startup peer with a rejected first close, and a post-close entry with the requested ID but a different input path. ### Authority boundaries and non-goals Diagnostics only. Nothing here closes, kills or adopts upstream sessions, and successful-close validation is not weakened. ### Alternatives considered Folding these into #1841. They were deferred because each review round in that PR raised further diagnostic edge cases, while its core fix is the CLI exit. ### Possible direction Record the startup `idb_list` same-input entries through the same blocker-recording path used for recovery and post-close inventories, and carry `input_path` in `UpstreamSession`. ### References #1841, #1710

bsakaguchi

2026-10-15 18:27:29 +00:00

closed this issue
added the
area: native

enhancement

P2: medium
labels

bsakaguchi commented

2026-10-15 18:27:30 +00:00

Author

Owner

@matbrgz commented on GitHub (Oct 12, 2026):

Diagnostics & Upstream Session Tracking Review

Completing upstream_sessions evidence in headless IDA cleanup_incomplete errors aligns with REA's core rule: never drop parsed ownership facts.

Inventory Correlation: If idb_list during startup identifies pre-existing sessions for the same input target, retaining them in the session tracker allows cleanup_incomplete to distinguish between sessions spawned by this run versus external or orphaned peer sessions.
Explicit Provenance: Emitting owned: true for the active session and owned: false (or external_peer) for pre-existing sessions gives callers clear actionable signals on which IDA database locks must be cleaned up manually.

@matbrgz commented on GitHub (Oct 12, 2026): ### Diagnostics & Upstream Session Tracking Review Completing `upstream_sessions` evidence in headless IDA `cleanup_incomplete` errors aligns with REA's core rule: never drop parsed ownership facts. - **Inventory Correlation:** If `idb_list` during startup identifies pre-existing sessions for the same input target, retaining them in the session tracker allows `cleanup_incomplete` to distinguish between sessions spawned by this run versus external or orphaned peer sessions. - **Explicit Provenance:** Emitting `owned: true` for the active session and `owned: false` (or `external_peer`) for pre-existing sessions gives callers clear actionable signals on which IDA database locks must be cleaned up manually.

bsakaguchi commented

2026-10-15 18:27:30 +00:00

Author

Owner

@morluto commented on GitHub (Oct 13, 2026):

Verified during the bug/fix-PR audit. Fixed on main by #2814 (commit 7369b5ee6). IDA retains successful startup inventory facts, ownership, input paths and conflicting bindings through failed cleanup.

Current implementation: src/ida/IdaSessionClient.ts. Validation on macOS arm64 with Node.js 24.18.0: npm run test:local -- src/application/HermesConfigurationYaml.test.ts src/ida/IdaSessionClient.test.ts src/native/NativeMachoInspection.test.ts src/browser/PlaywrightElectronActiveProvider.test.ts passed (84 tests). This is source/provider-boundary regression coverage; no real IDA GUI session was launched for this check.

Closing because the reported behavior is already addressed on main.

@morluto commented on GitHub (Oct 13, 2026): Verified during the bug/fix-PR audit. Fixed on main by #2814 (commit 7369b5ee6). IDA retains successful startup inventory facts, ownership, input paths and conflicting bindings through failed cleanup. Current implementation: `src/ida/IdaSessionClient.ts`. Validation on macOS arm64 with Node.js 24.18.0: `npm run test:local -- src/application/HermesConfigurationYaml.test.ts src/ida/IdaSessionClient.test.ts src/native/NativeMachoInspection.test.ts src/browser/PlaywrightElectronActiveProvider.test.ts` passed (84 tests). This is source/provider-boundary regression coverage; no real IDA GUI session was launched for this check. Closing because the reported behavior is already addressed on main.

bsakaguchi referenced this issue

2026-10-16 21:15:17 +00:00

[PR #698] [MERGED] fix(comparison): retain container-only export shape paths #2178

bsakaguchi added

enhancement

and removed

enhancement

labels

2026-10-31 00:29:27 +00:00