Artifact store judges a held version by the artifact's own digest; a failed build step keeps its error's head (H-A-474, H-A-475) #739

Merged
mesh-tools merged 2 commits from fix/held-digest into main 2026-11-22 14:01:32 +00:00
Member

Design lines

Design 47 H2, lead answers H-A-474 and H-A-475 (hq branch design/47-H1-answers-59, commit 33ba1815); H-A-393.

What changed

  • distribution (the artifact store): OCI.Held compares like with like per kind: a bundle's or an archive's stored layer digest, an image's manifest digest. The same digest is already held (the publish ends, published is said, nothing is stored again), on a repeated ask too; another digest is refused naming the held one.
  • mesh-build-node and mesh-build-image: a failed step's output keeps its head and its tail within the bound, with … <n> bytes left out … between them (the Node module's build step and toolchain install, the image module's reason and journal line). The Go build module already keeps a failure's first line and is unchanged.

Why

  • gitea's build was refused three times at identical bytes: Held compared the version tag's manifest digest with a bundle's blob digest, so every second publish of such a version was refused.
  • esbuild's own error line was lost: only the tail of stderr was kept, and the Node launcher's stack trace pushed it out ("esbuild for code: status 1" with no cause).

Tests

Unit tests in modules/distribution/internal/holder (a fake store keeping a manifest digest beside the layer's; a row running the module's OCI code over an in-memory store: same digest held, other digest refused, repeated ask), and in mesh-build-node and mesh-build-image (build_test.go: head and tail kept, the cut marked). Run by the developer and read by the reviewer at the approved head; the pull request's check runs them again.

Reviewed-by: mesh-code-reviewer@shanks at fb71ff5439
Pushed-by: lead-47@shanks

## Design lines Design 47 H2, lead answers H-A-474 and H-A-475 (hq branch design/47-H1-answers-59, commit 33ba1815); H-A-393. ## What changed - **distribution** (the artifact store): `OCI.Held` compares like with like per kind: a bundle's or an archive's stored layer digest, an image's manifest digest. The same digest is already held (the publish ends, published is said, nothing is stored again), on a repeated ask too; another digest is refused naming the held one. - **mesh-build-node** and **mesh-build-image**: a failed step's output keeps its head and its tail within the bound, with `… <n> bytes left out …` between them (the Node module's build step and toolchain install, the image module's reason and journal line). The Go build module already keeps a failure's first line and is unchanged. ## Why - gitea's build was refused three times at identical bytes: Held compared the version tag's manifest digest with a bundle's blob digest, so every second publish of such a version was refused. - esbuild's own error line was lost: only the tail of stderr was kept, and the Node launcher's stack trace pushed it out ("esbuild for code: status 1" with no cause). ## Tests Unit tests in `modules/distribution/internal/holder` (a fake store keeping a manifest digest beside the layer's; a row running the module's OCI code over an in-memory store: same digest held, other digest refused, repeated ask), and in `mesh-build-node` and `mesh-build-image` (`build_test.go`: head and tail kept, the cut marked). Run by the developer and read by the reviewer at the approved head; the pull request's check runs them again. Reviewed-by: mesh-code-reviewer@shanks at fb71ff5439f865a3e0d5a703d1f5f5fbcd0386e7 Pushed-by: lead-47@shanks
mesh-tools added 2 commits 2026-11-22 14:01:24 +00:00
Held compared the version tag's manifest digest with a bundle's or an
archive's blob digest, so every second publish of such a version was
refused even at the same bytes. It now compares like with like: a bundle's
or an archive's stored layer, an image's manifest. The same digest is
already held (ended, published said, nothing stored), on a repeated ask too;
another digest is refused naming the held one. The fake store keeps a
manifest digest beside the layer's, and a row runs the module's OCI code
over a store in memory.

Builds: design 47 H2 H-A-474, design 47 H H-A-393
The Node build module kept only the last half of its bound of a failed
command's stderr, so esbuild's error line was pushed out by its Node
launcher's stack trace ("esbuild for code: status 1" with no cause). The
image build module's reason and journal line kept stderr's tail alone too,
and so did the Node module's toolchain install. Each now keeps the head and
the tail within its bound, with "… <n> bytes left out …" between them. The
Go build module keeps a failure's first line, not its tail, and is left as
it is.

Builds: design 47 H2 H-A-475
mesh-tools merged commit c791e5aff2 into main 2026-11-22 14:01:32 +00:00
mesh-tools deleted branch fix/held-digest 2026-11-22 14:01:32 +00:00
Sign in to join this conversation.
No Reviewers
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: novox/mesh-catalog#739