Make Scoop digest helper fetch raw attachment bodies #1431

New issue

Closed

opened 2026-07-15 12:34:20 +00:00 by coilyco-ops · 2 comments

coilyco-ops commented

2026-07-15 12:34:20 +00:00

Member

Problem

PR #1429 closed #1420, but the post-merge release run still fails in bump-scoop-manifest with release asset metadata JSON where the Scoop job needs the raw 64-hex .sha256 body.

Run evidence:

run 8547 (#2266) failure release.yml push release "Make Scoop release digest helper return raw sidecar body (#1429)"
job bump-scoop-manifest
::error::sidecar digest '{"id":4316,"name":"ward-windows-amd64.exe.sha256","size":65,"download_count":0,"created_at":"2026-07-15T12:32:23Z","uuid":"c2a3d796-517a-44be-86b7-cfdfeacb2061","browser_download_url":"https://forgejo.coilysiren.me/attachments/c2a3d796-517a-44be-86b7-cfdfeacb2061","type":"attachment"}' is not a 64-hex sha256; refusing to write manifest
Job bump-scoop-manifest failed

PR #1429 added a metadata fallback in scripts/forgejo-release-asset.sh, but release still receives metadata JSON for the sidecar digest. The failing asset has size:65, which suggests the raw attachment body may be the desired digest plus newline, but the current job path still returns the Forgejo attachment metadata envelope.

This is a missed acceptance follow-up for #1420 / #1412. The release path still does not prove that bump-scoop-manifest reads raw .sha256 bodies after the #1416 checkout fix.

Acceptance

Make bump-scoop-manifest read the raw 64-hex .sha256 body for both Windows sidecar assets in the actual release job after #1429.
Preserve #1414 best-effort behavior for missing Scoop token / bucket push side effects.
Add or update contract coverage so the helper catches the current failure mode where the fallback URL still yields attachment metadata JSON instead of the digest.
Confirm release.yml succeeds for the current release branch SHA including bump-scoop-manifest, or leave exact blocker evidence if an external service prevents confirmation.

Related: #1420, #1429, #1412, #1416, run 8547 / #2266.

## Problem PR #1429 closed #1420, but the post-merge release run still fails in `bump-scoop-manifest` with release asset metadata JSON where the Scoop job needs the raw 64-hex `.sha256` body. Run evidence: ```text run 8547 (#2266) failure release.yml push release "Make Scoop release digest helper return raw sidecar body (#1429)" job bump-scoop-manifest ::error::sidecar digest '{"id":4316,"name":"ward-windows-amd64.exe.sha256","size":65,"download_count":0,"created_at":"2026-07-15T12:32:23Z","uuid":"c2a3d796-517a-44be-86b7-cfdfeacb2061","browser_download_url":"https://forgejo.coilysiren.me/attachments/c2a3d796-517a-44be-86b7-cfdfeacb2061","type":"attachment"}' is not a 64-hex sha256; refusing to write manifest Job bump-scoop-manifest failed ``` PR #1429 added a metadata fallback in `scripts/forgejo-release-asset.sh`, but release still receives metadata JSON for the sidecar digest. The failing asset has `size:65`, which suggests the raw attachment body may be the desired digest plus newline, but the current job path still returns the Forgejo attachment metadata envelope. This is a missed acceptance follow-up for #1420 / #1412. The release path still does not prove that `bump-scoop-manifest` reads raw `.sha256` bodies after the #1416 checkout fix. ## Acceptance - Make `bump-scoop-manifest` read the raw 64-hex `.sha256` body for both Windows sidecar assets in the actual release job after #1429. - Preserve #1414 best-effort behavior for missing Scoop token / bucket push side effects. - Add or update contract coverage so the helper catches the current failure mode where the fallback URL still yields attachment metadata JSON instead of the digest. - Confirm `release.yml` succeeds for the current release branch SHA including `bump-scoop-manifest`, or leave exact blocker evidence if an external service prevents confirmation. Related: #1420, #1429, #1412, #1416, run 8547 / #2266.

coilyco-ops added this to the warded control plane coherence milestone

2026-07-15 12:34:20 +00:00

coilyco-ops added the

headless

labels

2026-07-15 12:34:20 +00:00

coilyco-ops commented

2026-07-15 12:34:40 +00:00

Author

Member

DISPATCH-DEFERRED: capacity blocked. Attempted warded engineer coilyco-flight-deck/ward#1431 --harness codex --workflow pull-request-and-merge at 2026-07-15T12:34Z from the director surface. Ward rejected the launch because coilyco-flight-deck/ward already has 3 active engineers at the repo limit. Broker evidence: repo engineer limit is reached for coilyco-flight-deck/ward: 3 active engineer(s) (limit 3). Dispatch log: /Users/kai/.ward/agent-logs/dispatch/20260715T123426Z-director-codex-jp85-coilyco-flight-deck-ward-1431.log. Retry this remaining release/Scoop digest blocker as soon as a slot opens.

DISPATCH-DEFERRED: capacity blocked. Attempted `warded engineer coilyco-flight-deck/ward#1431 --harness codex --workflow pull-request-and-merge` at 2026-07-15T12:34Z from the director surface. Ward rejected the launch because coilyco-flight-deck/ward already has 3 active engineers at the repo limit. Broker evidence: repo engineer limit is reached for coilyco-flight-deck/ward: 3 active engineer(s) (limit 3). Dispatch log: /Users/kai/.ward/agent-logs/dispatch/20260715T123426Z-director-codex-jp85-coilyco-flight-deck-ward-1431.log. Retry this remaining release/Scoop digest blocker as soon as a slot opens.

coilyco-ops commented

2026-07-15 12:35:26 +00:00

Author

Member

DUPLICATE/SUPERSEDED: #1430 was already filed and dispatched before this issue, and it is actively running as engineer-codex-ward-1430 for the same post-#1429 release failure (run 8547 / #2266). This issue kept the identical-label #1420 follow-up evidence, but the repair lane should continue through #1430 to avoid duplicate engineers.

DUPLICATE/SUPERSEDED: #1430 was already filed and dispatched before this issue, and it is actively running as `engineer-codex-ward-1430` for the same post-#1429 release failure (run 8547 / #2266). This issue kept the identical-label #1420 follow-up evidence, but the repair lane should continue through #1430 to avoid duplicate engineers.

coilyco-ops closed this issue

2026-07-15 12:35:26 +00:00