[Clear1nsight]

Change and mathematical scope

Implements the concrete pure enabling component returned by the completed compiler-input closure Pro consultation: exact six-source root/argv construction and a conditional ordinary-header no-PCH first-match prefix. It authenticates all six original role C/header occurrences, emits all 20 literal include obligations and preserves the unchanged list-valued GCC command. Independent executable controls retain corruption/no-op distinctions, negative-lookup and search-order cases, unsupported cases and a stale-facts counterexample.

Partial compiler-input closure prerequisite only. The function does not derive the installed compiler's effective search chain, authenticate supplied namespace facts, construct a genuine observer, or complete macro/specs/linker/loader closure. Full compiler-input closure and the bootstrap-control campaign, original controls, actual namespace/currentness, runtime qualification, complete costs and whole-activity credit remain open. A hash-bound report does not prove its assertions true.

Task and overlap

Original source investigation: PR #6321. Frozen H25 benchmark: PR #6029. Immediate predecessor: PR #6352.

Standing explicit user assignment: consult Pro, implement useful prerequisites and contribute after completed units; latest instruction publishes resolved questions and requests the compiler-input closure consultation. This is the accurately named new enabling unit from that completed response, not a resumed bootstrap-control campaign round or a narrowed full-family claim.

• Canonical base 7c49fe32c55de02c2c873655c804b16e4c3fef0a.
• Public head b3771dbc5f713ba3e90170c1dc3321e653222c65; checked tree 09bd5dc1ed51907fcac58789533a949c67a6e4bb.
• Nineteen new files under research/h25-c05-source-roots/, including nine exact inert functional inputs for cold availability. No unmerged executable dependency or unavailable original Git ancestry is assumed.
• Companion Clarify caught-primary preservation and accounting requirements #6351 1e9db2d08be4551ffe0bcfc573574851530eb4ca clarifies G12 original scope; Resolve bootstrap entry, preflight and dependency source questions #6352 b280be8dab15359159538e639b1d8c7deafc0612 publishes preceding bootstrap-control campaign source resolutions. This PR adds executable pure enabling bodies and independent controls, rather than repeating those audits. Original lineage Archive H25 cache sources and remaining N22, P07 and P02 work #6321/Reduce complete H25 bookkeeping time by 21% #6029/Reduce complete precision qualification time by 19% with C bookkeeping #5864 and separate Cursor/h25 g12 identity cleanup 5d9d #6339/Repair bounded G12 custody cleanup and publication-size checks #6346/Record G12 acquisition and raw-error delivery boundaries #6347 scopes remain.
• The 6365-byte implementation exactly matches Pro's returned SHA256 438cdc6c8b003d9a1b6a41889d6ac2f3d5bed25f77b430cf3b5f63c36c4e4612. Codex did not alter it. Pro reports a separate agent review with repairs; Codex independently reviewed the returned source and constructed its own controls.
• The exact original helper at SHA256 aa0a315110e031e19291970c9611fe5e0f91465fa20eaec51e81186099563f78 is newly retained as inert text. Its sole Popen call supplies neither env nor cwd; actual inherited caller context remains unqualified. This input was found after the completed Pro response.
• Scope changed from the retained portable-replay card to the explicit fixed-source compiler-input closure enabling assignment. The original source/configuration/gates are preserved.

Validation

Executed on Python 3.12.3:

python3 -B scripts/research_package.py --package research/h25-c05-source-roots --check-only
python3 -B scripts/research_package.py --package research/h25-c05-source-roots
python3 -B scripts/check_research_commit.py --package research/h25-c05-source-roots

Metadata/pins and registered normal/optimized replay pass. Each mode reports controls=53 project_paths=6 literal_sites=20 stale_facts_counterexample=1 full_C05=false runtime_qualified=false. Independent expected argv is derived from the original producer AST, and the site oracle separately enumerates all fixed-source literal directives. Controls cover all six byte corruptions, distinct-header swap refusal, equal-content positives, exact argument/path/mode domains, missing earlier observations, earlier winners, search-order changes, unsupported PCH/alias/error outcomes, malformed unused suffixes and the preserved stale-data limitation.

The 19-file committed-byte preflight passed. Exact remote branch, tree and sole canonical parent are verified. Hosted checks are pending at submission. A metadata preparation refusal for the required Open next questions heading was corrected without changing implementation, fixtures or controls; no functional source repair was needed.

Only the new pure implementation is executed. Archived producer/helper are parsed as AST and original C/header payloads are inert bytes; no Root/provider/producer/helper execution, native build, compiler diagnostic, original control or real namespace observer is run. The finite namespace inputs are explicitly data, not a fake Root/owner/syscall observation.

DOMAIN-ARGUMENT.md states the conditional induction and all external premises. Supplied mode labels and required 0444 values are not flag/stat observations. The new selection markers remain None and are not guards preventing pure testing; original execution/admission gates are untouched. Local component bounds do not redefine compiler-input closure or increase original limits. Unknown physical/preparation/review/publication/final costs remain OPEN_NOT_ZERO.

Credit

Chris R / Clear1nsight with substantial ChatGPT Pro construction/reported agent-review and OpenAI Codex independent source/control/publication assistance. Retain original source/reviewer and PR6321/6029/5864 attribution, unpaidmoon/Cursor Grok for #6339, and Douglas R. Colkitt's circuit/network/native/verification lineage. Snapshots/notices unchanged; new source/controls/analysis Apache-2.0. AI review and finite cases are not human or proof-kernel acceptance.

Open next questions

Bind installed toolchain configuration/source correspondence and actual caller context, then implement complete applicable root/expansion rules and a genuine namespace/alias/currentness observer with original custody/first-error guarantees. Establish selected parent record bounds and actual total fit/cost. Existing source roots and a conditional prefix certificate do not discharge those edges. See OPEN-EDGES.json, DOMAIN-ARGUMENT.md and TASK.json. No automatic native preparation, new trust premise or prior-round restart is authorised by this contribution.

[Clear1nsight]

Publication readback complete at exact head b3771dbc5f713ba3e90170c1dc3321e653222c65, tree 09bd5dc1ed51907fcac58789533a949c67a6e4bb: one commit, 19 intended new files, canonical base 7c49fe32c55de02c2c873655c804b16e4c3fef0a, draft state and maintainer edits disabled. Exact title/body and all remote file blobs match the frozen local package.

Hosted run 41387009542 completed attempt 1 successfully: plan job 125089953546 and actual bounded replay job 125091142576, including unchanged committed inputs. Local metadata/pins, registered normal/optimized 53-control runs and the 19-file committed-byte preflight passed.

The canonical completion handoff records the completed Pro consultation, separate #6351/#6352 source resolutions, exact findings and remaining dependencies. This package constructs six pinned project path occurrences and 20 literal sites; its ordinary no-PCH prefix result is conditional on authentic, stable, correctly interpreted namespace facts. The preserved stale-facts counterexample prevents treating supplied data as a real observer.

Full compiler-input closure and bootstrap-control campaign remain open. Installed configured rules/source correspondence, actual caller context, complete expansion/observer, parent limits, actual complete fit and costs remain unresolved. Only the new pure component and scoped source verifiers ran. Original Root/provider/helper/compiler/native/scientific work did not run. Complete physical costs remain OPEN_NOT_ZERO. The bounded unit is delivered; it does not restart the stopped bootstrap-control campaign round.

Original H25 benchmark: PR #6029. Original source investigation: PR #6321.