Repository navigation
🐛🧑💻 - Support complete large comparison assessment after file recovery budget exhaustion in #773
Description
Is there an existing issue for this?
- I have searched the existing issues.
Which actions are affected?
Commits
Which platforms are affected?
Linux
Description
Copilot 3.10.6 includes #726, but the same coherent consumer specification package still cannot obtain a progress assessment. The failure has changed from missing comparison patches to validation.evidence-too-large: file recovery budget exhausted, with all 140 paths reported unreviewed. Progress is correctly retained and no assessment is published. Please provide a supported complete, bounded assessment path for this workload, without making consumers reduce product scope, split a coherent specification package around an implementation limit, or weaken evidence requirements.
This is a remaining capability gap after the released #726 fix, rather than a claim that fail-closed behavior should be removed. Preserve per-request and total resource controls. A resumable/multipart evidence and assessment protocol must bind every part to one immutable base/head, detect incomplete/stale/replayed parts, and publish progress only after complete reviewed coverage.
Reproducing the issue
- Consumer PR: https://github.com/vypdev/geckocv/pull/32 (private repository; access required).
- Exact head:
55792600c0be6aa581d34880073eb29abde634b7. - Run: https://github.com/vypdev/geckocv/actions/runs/40764178299/job/123083764814.
- Setup resolves
vypdev/copilot@v3tof5e5a4cf9a80248994469cd88ad55eecab5b95ca, package version 3.10.6. - Progress metadata pins base
b461b78c035928856862b995d9008a64b3d59848and that exact head. The PR was originally branched at2ef0f5d5125b08e2a481d30ede0e1cdd88e4cc85; retain the run's actual assessment base when reproducing. - The run reports file recovery budget exhaustion and 140 unreviewed paths before model assessment. The PR metadata lists 140 changed files and 72,064 additions against its original merge base. It contains 100 English SDDs, granular traceability/verification registries, documentation and repository quality checks, with no binary evidence requirement.
- Specification and architecture CI succeeds for this exact head. Its 96 local Node fixtures and six parsed-workflow fixtures are separate verification evidence, not a substitute for the blocked assessment.
copilot Version
3.10.6; vypdev/copilot@v3 resolved to f5e5a4cf9a80248994469cd88ad55eecab5b95ca in the failing run.
Relevant Log Output
Progress was not assessed: file recovery budget exhausted. Unreviewed paths (140): ".github/scripts/tests/test_specification_workflow.py", ".github/workflows/docker_client_contract.yml", ".github/workflows/specifications.yml", "AGENTS.md", "README.md", "docs/audits/2026-10-26-full-scope-specifications.md", "docs/audits/2026-10-26-functional-baseline.md", "docs/audits/2026-10-26-specification-package.md", ... Existing progress was retained.
Error code: validation.evidence-too-large
Retryable: No
Reference: 1af78c4f-fefe-4b8c-89b3-9c4fc360df77
Additional context and comments
The maintainer explicitly requests reporting Copilot limitations upstream and retaining the entire approved specification scope. #726 is closed and included in v3.10.6; this report captures the distinct post-release recovery-budget boundary. No gate, evidence completeness rule, limit, workflow version or consumer scope was changed to bypass the failure.
Suggested acceptance: exercise a realistic large coherent text-only comparison through the complete supported protocol; retain bounded resource use per part/run; track required paths and reviewed coverage at immutable SHAs; recover omitted patches with exact contents; fail closed for missing or unsupported content; detect moved heads, corruption and replay; aggregate only complete evidence; clearly report resource usage, uncovered paths and an actionable supported continuation route. Cover rejection and successful completion, rather than only raising a global byte/file constant or silently excluding the large specification registry.
Metadata
Metadata
Assignees
Labels
Type
Projects
- Status
Activity
[-]Support complete large comparison assessment after file recovery budget exhaustion in 3.10.6[/-][+]🐛 - Support complete large comparison assessment after file recovery budget exhaustion in[/+]vypbot commentedon Oct 28, 2026
Hi! I’m @vypbot, the Copilot assistant for this repository.
I can answer questions, explain the codebase, propose implementation and test plans, review issues and pull requests for potential bugs or security problems, and help authorized maintainers apply changes.
Try
/copilot helpto see the available commands, or mention @vypbot with your question.[-]🐛 - Support complete large comparison assessment after file recovery budget exhaustion in[/-][+]🐛🧑💻 - Support complete large comparison assessment after file recovery budget exhaustion in[/+]3 remaining items
vypbot commentedon Oct 29, 2026
Issue closed — historical progress
Current status: closed as completed.
The last estimate is retained below as history. Closure does not certify a new percentage, tests, or branch alignment.
Action required: none for this closed issue. If work continues, reopen the issue; a fresh assessment can update this same card.
Last assessment before closure (historical)
Merge pull request #790 from vypdev/bugfix/773-support-complete-large…
vypbot commentedon Oct 29, 2026
Issue closed: synchronization recommendation retired
No branch synchronization is required for this closed issue.
The previous recommendation is historical. Issue closure does not certify that the branches are synchronized.
efraespada commentedon Oct 29, 2026
Consumer old-release observation, not a post-fix regression:
0ac17237553237954d13c9b5ff7d464177e140b8, assessment base5dde8907973d3e9718c9e5886daa8a6caf7ae078.vypdev/copilot@v3atf5e5a4cf9a80248994469cd88ad55eecab5b95ca(3.10.6), which predates this issue's merged fix.validation.evidence-too-large,Retryable: No, reference2e2e76f7-17b0-471c-aec9-132b3e7cb02c. Existing progress is retained. No assessment/model-result validity is inferred.No retry, issue reopening, duplicate report, release/tag movement or scope reduction is requested from this old-release result. Consumer verification of the merged multipart fix requires an execution that actually includes it. This comment records the release/adoption boundary and a fresh reproducible workload while preserving strict evidence completeness.
The next documentation checkpoint
2d5ee72e19be24ccb416801b2a41693662382c8creproduces the same old-release boundary in run https://github.com/vypdev/geckocv/actions/runs/40960546078, attempt 1, base still5dde8907973d3e9718c9e5886daa8a6caf7ae078. Setup again downloads v3 atf5e5a4cf9a80248994469cd88ad55eecab5b95ca. At 21:47:22Z progress is not assessed: recovery budget exhausted, 146 unreviewed paths, primaryvalidation.evidence-too-large, nonretryable, reference71f7aa74-78f2-45f8-8585-e6e60ffc4d01; prior progress is retained. Exact-head specification/architecture CI https://github.com/vypdev/geckocv/actions/runs/40960557728 passes. All 100 SDDs/969 requirements/994 proposed cases remain intact. No retry, reopening, scope reduction or claim about the merged fix is requested.Further old-release observation, not a post-fix regression: documentation source
1a6676a30780d057e556038e85a5974704e1022c(GeckoCV PR #32), run https://github.com/vypdev/geckocv/actions/runs/40966495220, attempt 1. Setup explicitly resolvesvypdev/copilot@v3tof5e5a4cf9a80248994469cd88ad55eecab5b95ca(3.10.6). Progress is not assessed because file recovery budget is exhausted, with 147 paths unreviewed; primaryvalidation.evidence-too-large, retryable No, reference7f788595-ed60-463e-b496-bd83ddcf469a. Prior progress is retained. The exact-head specification/architecture CI passes in https://github.com/vypdev/geckocv/actions/runs/40966506765; local checks retain all 100 SDDs, 969 requirements and 994 proposed cases with 96 Node methods passing. This source records an external unmerged implementation receipt without fabricating baseline source adoption or completed product acceptance. No retry, reopening, scope reduction, limit increase or gate bypass is requested; the merged upstream multipart fix still needs an actual consuming release/run.Additional old-release observation: documentation source
d42a150f49f37661e2dd85774e76de2c564fae19, run https://github.com/vypdev/geckocv/actions/runs/40970660001, attempt 1, explicitly resolves v3 tof5e5a4cf9a80248994469cd88ad55eecab5b95ca/ 3.10.6. Evidence recovery budget is exhausted before assessment, with 148 paths unreviewed, nonretryablevalidation.evidence-too-large, reference284cd626-61eb-43ed-b89f-ab3cc722edaa. Prior progress is preserved. Exact-head specification/architecture CI https://github.com/vypdev/geckocv/actions/runs/40970672429 passes; all 100 SDDs, 969 requirements and 994 proposed cases retain scope/status. The checkpoint records external candidate evidence without fabricating source adoption or full API acceptance. This does not exercise the merged fix; no retry/reopening, scope reduction or gate/limit change is requested.Further old-release observation: documentation source
0c6164ee0e4e2a926952aeacf4b3879cc28e241a, run https://github.com/vypdev/geckocv/actions/runs/40975150767, attempt 1. Setup explicitly resolves v3 tof5e5a4cf9a80248994469cd88ad55eecab5b95ca/ 3.10.6. Recovery budget exhaustion leaves 149 paths unreviewed before assessment;validation.evidence-too-large, retryable No, reference02cad695-9a0b-40b7-8648-3337ca606558. Prior progress is retained; optional Checks 403 is secondary. Exact-head specification/architecture CI https://github.com/vypdev/geckocv/actions/runs/40975159201 passes. This immutable checkpoint retains 100 SDDs, 969 requirements and 994 proposed cases, plus separately scoped external implementation receipts. It does not exercise the merged fix; no blind retry, reopening, scope reduction or gate/limit change is requested.efraespada commentedon Oct 30, 2026
Another retained consumer observation on the unchanged published old runtime, not a regression claim against the merged develop fix:
0398825b5ef32017d9e1169f324d5807e6ac5296.vypdev/copilot@v3tof5e5a4cf9a80248994469cd88ad55eecab5b95ca(3.10.6).validation.evidence-too-large; referencef7502c7b-766d-4886-9f5f-28d1dc53b2f1. Existing progress is retained rather than a percentage being fabricated.35af10e52be7a6a9a09c19ff2ee6d1b81aa9aed2with verified parentsf6c43676e05a694b627571172356f8e26f8d7efcand the source above. Backend, frontend and PR-review workflows also finish successfully; unresolved substantive review is still handled separately.No unchanged-state retry, specification shrinkage, threshold reduction, native check fabrication, upstream branch or release was requested. This supplies another exact run for validating the already owned fix when it is published. The consumer has not executed an unverified newer release, and this comment does not imply that the full product or all its proposed cases are implemented.