morluto/rea

Fork 0

mirror of https://github.com/morluto/rea.git synced 2026-11-05 06:45:22 +00:00

[GH-ISSUE #4190] [Feature] Query retained application metadata and native facts by exact path #1642

New issue

Closed

opened 2026-10-16 21:08:21 +00:00 by bsakaguchi · 2 comments

bsakaguchi commented

2026-10-16 21:08:21 +00:00

Owner

Originally created by @N0zoM1z0 on GitHub (Oct 15, 2026).
Original GitHub issue: https://github.com/morluto/rea/issues/4190

Before submitting

I searched open and closed issues and PRs for related proposals.
I understand contribution scope must be reviewed before integration.
This request contains no credentials, private samples or sensitive target data.

Area

Evidence, artifacts, or reconstruction

Current limitation

A narrow question about retained application metadata should not require a whole reachable feature graph: which selected package/HTML/source-map/native files are present, which processing phases remain unavailable, and what supports those facts?

Executed on main c7f6b8bad1a34d5efb30ff2a017672dea1fcdabb, Linux x64, Node 24.18.0/npm 11.16.0, official Joplin v3.7.21 ASAR. After summary-first analysis of the paths below, an exact package.json outgoing trace completed but its response exceeded the stdio budget: at least 10,587,373 bytes, against a 10,484,736-byte result budget. The trace remains retained as ev_64903e99c2a09d15abd0955b8ddcab0dd4fb01b5a6721c64d2308b3a105df5ac. A second run reproduced the same parent/trace IDs and limits. This is a failed delivery, not a successful equivalent task or performance improvement.

{"name":"analyze_javascript_application","arguments":{"input_path":"/absolute/path/to/app.asar","detail":"summary","source_paths":["ElectronAppWrapper.js","ElectronAppWrapper.js.map","index.html","package.json","node_modules/keytar/build/Release/keytar.node"]}}
{"name":"trace_application_feature","arguments":{"application":{"kind":"retained-evidence","evidence_id":"<returned parent_evidence_id>"},"native_observations":[],"seed":{"kind":"string","value":"package.json","match":"exact","case_sensitive":true},"direction":"outgoing"}}

The published application was not executed. Acquisition/extraction and input digests are in the Joplin verification record. The buffer limit is useful; the missing piece is a suitable bounded fact query.

Use cases

Resume retained Electron analysis and inspect several known metadata/native paths, including after a source timeout. #3948 owns incorrect processing applicability; this request owns delivery of the resulting facts. No new code-mode executor is proposed (#4091).

Desired outcome and success criteria

Reuse inspect_analysis_view and the same CLI/MCP projection owner. Read existing parsed facts without rerunning or reinterpreting source.
Select exact paths/kinds and needed fact fields; support the homogeneous multi-path question above and bounded paging with stable ordering.
Return canonical node identity, precise retained locations, selected metadata and applicable processing gaps inline, with parent Evidence, source basis, coverage and partial/unknown states. Preserve multiple locations rather than choosing an arbitrary display path.
Keep complete Evidence accessible. Zero selected rows must not imply absence outside examined coverage.
Verify equivalent completed real-app workflows before/after, including schemas, requests, replies, extra calls/retries, elapsed/CPU time and peak process-family memory; distinguish first/repeated queries, uncontrolled caches and unmeasured host/model costs. Overflow is not an efficiency baseline.

Expected interaction with targets

Read existing retained local Evidence. Existing acquisition effects stay with their current tools; this view adds no target execution, injection or mutation.

Verification available so far

I have observed results. Proposed view correctness, cost and runtime behavior: NOT_ENOUGH_INFO — not implemented. Keep reproducible scripts and compact measurements; do not archive downloadable applications or large Evidence.

Alternatives considered / possible direction

#4092/#4097 provide useful typed module references, but their owning contract covers javascript-asset, javascript-module and source-module. Generic module item selectors share that scope. Full feature tracing answers a broader relationship question. Exporting complete retained Evidence and selecting facts locally remains a recovery path and a candidate completed baseline.

Extend the existing retained-view contracts with a focused application-node/fact projection after checking current owners and clients. #1050 is the original selected-view work; #3134 measures retained-history paging and does not own this representation change. Part of #4027 and #1065. No implementation or integration approval is implied by this proposal.

Originally created by @N0zoM1z0 on GitHub (Oct 15, 2026). Original GitHub issue: https://github.com/morluto/rea/issues/4190 ### Before submitting - [x] I searched open and closed issues and PRs for related proposals. - [x] I understand contribution scope must be reviewed before integration. - [x] This request contains no credentials, private samples or sensitive target data. ### Area Evidence, artifacts, or reconstruction ### Current limitation A narrow question about retained application metadata should not require a whole reachable feature graph: which selected package/HTML/source-map/native files are present, which processing phases remain unavailable, and what supports those facts? Executed on main `c7f6b8bad1a34d5efb30ff2a017672dea1fcdabb`, Linux x64, Node 24.18.0/npm 11.16.0, official Joplin v3.7.21 ASAR. After summary-first analysis of the paths below, an exact `package.json` outgoing trace completed but its response exceeded the stdio budget: at least **10,587,373 bytes**, against a **10,484,736-byte result budget**. The trace remains retained as `ev_64903e99c2a09d15abd0955b8ddcab0dd4fb01b5a6721c64d2308b3a105df5ac`. A second run reproduced the same parent/trace IDs and limits. This is a failed delivery, not a successful equivalent task or performance improvement. ```json {"name":"analyze_javascript_application","arguments":{"input_path":"/absolute/path/to/app.asar","detail":"summary","source_paths":["ElectronAppWrapper.js","ElectronAppWrapper.js.map","index.html","package.json","node_modules/keytar/build/Release/keytar.node"]}} {"name":"trace_application_feature","arguments":{"application":{"kind":"retained-evidence","evidence_id":"<returned parent_evidence_id>"},"native_observations":[],"seed":{"kind":"string","value":"package.json","match":"exact","case_sensitive":true},"direction":"outgoing"}} ``` The published application was not executed. Acquisition/extraction and input digests are in the [Joplin verification record](https://github.com/morluto/rea/blob/c7f6b8bad1a34d5efb30ff2a017672dea1fcdabb/docs/verification/javascript-native-handoff-identity-2026-10-15.md). The buffer limit is useful; the missing piece is a suitable bounded fact query. ### Use cases Resume retained Electron analysis and inspect several known metadata/native paths, including after a source timeout. #3948 owns incorrect processing applicability; this request owns delivery of the resulting facts. No new code-mode executor is proposed (#4091). ### Desired outcome and success criteria - Reuse `inspect_analysis_view` and the same CLI/MCP projection owner. Read existing parsed facts without rerunning or reinterpreting source. - Select exact paths/kinds and needed fact fields; support the homogeneous multi-path question above and bounded paging with stable ordering. - Return canonical node identity, precise retained locations, selected metadata and applicable processing gaps inline, with parent Evidence, source basis, coverage and partial/unknown states. Preserve multiple locations rather than choosing an arbitrary display path. - Keep complete Evidence accessible. Zero selected rows must not imply absence outside examined coverage. - Verify equivalent completed real-app workflows before/after, including schemas, requests, replies, extra calls/retries, elapsed/CPU time and peak process-family memory; distinguish first/repeated queries, uncontrolled caches and unmeasured host/model costs. Overflow is not an efficiency baseline. ### Expected interaction with targets Read existing retained local Evidence. Existing acquisition effects stay with their current tools; this view adds no target execution, injection or mutation. ### Verification available so far I have observed results. Proposed view correctness, cost and runtime behavior: **NOT_ENOUGH_INFO — not implemented**. Keep reproducible scripts and compact measurements; do not archive downloadable applications or large Evidence. ### Alternatives considered / possible direction #4092/#4097 provide useful typed module references, but their owning contract covers `javascript-asset`, `javascript-module` and `source-module`. Generic module item selectors share that scope. Full feature tracing answers a broader relationship question. Exporting complete retained Evidence and selecting facts locally remains a recovery path and a candidate completed baseline. Extend the existing retained-view contracts with a focused application-node/fact projection after checking current owners and clients. #1050 is the original selected-view work; #3134 measures retained-history paging and does not own this representation change. Part of #4027 and #1065. No implementation or integration approval is implied by this proposal.

bsakaguchi

2026-10-16 21:08:21 +00:00

closed this issue
added the
triage

enhancement
labels

bsakaguchi commented

2026-10-16 21:08:22 +00:00

Author

Owner

@N0zoM1z0 commented on GitHub (Oct 15, 2026):

Implementation in progress on feat/javascript-retained-facts, based on latest main 85eac7b92bac39bf2da6c00ebf6ba1b11b4c8d1f after #4207 merged. No PR yet.

The prototype extends inspect_analysis_view with javascript-facts: exact retained location/kind filters, caller-selected property keys, and observation-level paging with canonical identities, original Evidence and source coverage. Shared module/fact coverage has one owner; CLI/MCP use the existing application service. No target execution or new executor.

Real Joplin completed export+local-filter and view answers agree for normal and 1 ms interrupted analysis (12/11 observations), with exact view CLI/MCP parity. View retrieval is much faster in these runs, but selected-catalog static task BPE is higher (normal 17,289 vs 8,887): the generic view schema and Evidence envelope outweigh the saved caller program. This is not a token-efficiency win. Actual fixed-model agent pilots are underway before deciding the final integration; preserve failed harness attempts and all resource/token receipts. Scripts and compact records will be retained, not downloadable applications or large exports.

@N0zoM1z0 commented on GitHub (Oct 15, 2026): Implementation in progress on `feat/javascript-retained-facts`, based on latest main `85eac7b92bac39bf2da6c00ebf6ba1b11b4c8d1f` after #4207 merged. No PR yet. The prototype extends `inspect_analysis_view` with `javascript-facts`: exact retained location/kind filters, caller-selected property keys, and observation-level paging with canonical identities, original Evidence and source coverage. Shared module/fact coverage has one owner; CLI/MCP use the existing application service. No target execution or new executor. Real Joplin completed export+local-filter and view answers agree for normal and 1 ms interrupted analysis (12/11 observations), with exact view CLI/MCP parity. View retrieval is much faster in these runs, but selected-catalog static task BPE is higher (normal 17,289 vs 8,887): the generic view schema and Evidence envelope outweigh the saved caller program. This is not a token-efficiency win. Actual fixed-model agent pilots are underway before deciding the final integration; preserve failed harness attempts and all resource/token receipts. Scripts and compact records will be retained, not downloadable applications or large exports.

bsakaguchi commented

2026-10-16 21:08:22 +00:00

Author

Owner

@N0zoM1z0 commented on GitHub (Oct 15, 2026):

Implemented in #4265 (ready for review), final local head 7238fd0802c399d9349d8f235b2526c57e5682a5, after integrating pinned main 2202e3f0ba556d40cc6e598df1ecbdf1eda56c13.

Existing CLI/MCP views now select exact retained application observations by batched paths/kinds and requested property keys, with optional property-name discovery. Original identity/provenance, missing versus null, page coverage and global partial/unknown states remain separate. Semantic relationships and unknown records remain accessible through complete Evidence/trace; no executor or persistent cache is added.

Executed: 101 focused cases across six files, full static/docs checks, five consistency cases, discovery, and fourteen version-labelled real Joplin export/view workflows. All oracle checks passed; all seven view cases had exact complete CLI/MCP Evidence parity. Final source hashes match; one CPU/8 GiB, no OOM and verified cleanup.

The cost result is mixed: integrated normal retrieval was 3.232 s export/projection versus 0.149 s view, but selected-contract structured BPE was 8,936 versus 17,393. Seven valid Terra/high Codex CLI trials plus one excluded host-configuration failure remain recorded. The partial-task candidate avoided export/shell and used less input, but more CPU/wall time; normal field discovery was not a stable win. No universal token/latency benefit or production code-mode recommendation is claimed.

Reproduction, exact versions, compact measurements/answers, failed attempts and limitations. Downloadable apps and large Evidence/transcripts are not archived. CI and final-head automated reviews are pending.

@N0zoM1z0 commented on GitHub (Oct 15, 2026): Implemented in #4265 (ready for review), final local head `7238fd0802c399d9349d8f235b2526c57e5682a5`, after integrating pinned main `2202e3f0ba556d40cc6e598df1ecbdf1eda56c13`. Existing CLI/MCP views now select exact retained application observations by batched paths/kinds and requested property keys, with optional property-name discovery. Original identity/provenance, missing versus null, page coverage and global partial/unknown states remain separate. Semantic relationships and unknown records remain accessible through complete Evidence/trace; no executor or persistent cache is added. Executed: 101 focused cases across six files, full static/docs checks, five consistency cases, discovery, and fourteen version-labelled real Joplin export/view workflows. All oracle checks passed; all seven view cases had exact complete CLI/MCP Evidence parity. Final source hashes match; one CPU/8 GiB, no OOM and verified cleanup. The cost result is mixed: integrated normal retrieval was 3.232 s export/projection versus 0.149 s view, but selected-contract structured BPE was 8,936 versus 17,393. Seven valid Terra/high Codex CLI trials plus one excluded host-configuration failure remain recorded. The partial-task candidate avoided export/shell and used less input, but more CPU/wall time; normal field discovery was not a stable win. No universal token/latency benefit or production code-mode recommendation is claimed. [Reproduction, exact versions, compact measurements/answers, failed attempts and limitations](https://github.com/morluto/rea/blob/7238fd0802c399d9349d8f235b2526c57e5682a5/docs/verification/javascript-retained-facts-2026-10-15.md). Downloadable apps and large Evidence/transcripts are not archived. CI and final-head automated reviews are pending.

bsakaguchi referenced this issue

2026-10-17 21:01:32 +00:00

[GH-ISSUE #2383] [Tracker] Audit analysis and debugging coverage for the TIOBE top 100 languages (October 2026) #794

bsakaguchi added

enhancement

and removed

enhancement

labels

2026-10-17 21:02:36 +00:00

bsakaguchi referenced this issue

2026-10-31 00:54:19 +00:00

[PR #1642] [CLOSED] fix(browser): decode prefixed source-map responses #4719