config: remove build binary field #372

Merged

caution-bot merged 0 commits from refs/pull/372/head into main

2026-07-13 21:36:30 +00:00

caution-bot commented

2026-07-10 02:24:04 +00:00

(Migrated from codeberg.org)

Summary

remove the build.binary field from caution.hcl parsing so new configs reject it instead of driving binary-only EIF builds
stop translating legacy binary: Procfile entries into caution.hcl and update templates/fixtures to rely on Containerfile/Dockerfile packaging
remove the remote builder binary_path request surface so hosted builds always emit full-image manifests without a binary entry

Closes #330

Test plan

cargo test -p caution-config
cargo test -p cli migrate_procfile
cargo test -p api builder
cargo check -p caution-config -p api -p cli
git diff --check

Notes:

The legacy EnclaveManifest.binary reader/build-helper path is intentionally left in place for previously produced manifests, but no current caution.hcl, Procfile migration, CLI local build path, or API remote build request can set build.binary/binary_path for new deployments.

## Summary - remove the `build.binary` field from `caution.hcl` parsing so new configs reject it instead of driving binary-only EIF builds - stop translating legacy `binary:` Procfile entries into `caution.hcl` and update templates/fixtures to rely on Containerfile/Dockerfile packaging - remove the remote builder `binary_path` request surface so hosted builds always emit full-image manifests without a `binary` entry Closes #330 ## Test plan - `cargo test -p caution-config` - `cargo test -p cli migrate_procfile` - `cargo test -p api builder` - `cargo check -p caution-config -p api -p cli` - `git diff --check` Notes: - The legacy `EnclaveManifest.binary` reader/build-helper path is intentionally left in place for previously produced manifests, but no current `caution.hcl`, Procfile migration, CLI local build path, or API remote build request can set `build.binary`/`binary_path` for new deployments.

caution-bot commented

2026-07-13 06:34:51 +00:00

(Migrated from codeberg.org)

E2E validation completed successfully on d7a5bdb587bbdad8f7ae29c40ad18fd73708156a.

Results:

make test-e2e: passed
- Happy path: 10 passed, 0 failed
- Platform/internal ports and encrypted routing: 8 passed, 0 failed
- Environment parity: 9 passed, 0 failed
- SSH-signed CLI operations: 9 passed, 0 failed
make test-e2e-builder: 10 passed, 0 failed
- Dedicated builder completed, EIF and DB metadata were verified, the builder terminated, and the repeated build hit the cache.
BYOC E2E: 9 passed, 0 failed
- caution init --byoc, customer-bucket builder deployment, attestation verification, and teardown all passed.

The make test-e2e-byoc wrapper stalled while waiting on its flock despite no visible lock holder, so I ran the target body directly with the same configured environment, setup, published provisioner, test script, cleanup trap, and teardown. The underlying BYOC test passed cleanly.

The happy-path suite reported its existing non-failing No HTTP Gateway Configured response warning. The same warning is present in earlier pre-branch E2E logs, so it does not appear to be introduced by this PR.

All provisioned test resources and local test containers were cleaned up. No regression was found in current-image builds, dedicated-builder builds, BYOC builds, deployment, attestation, environment handling, or platform-port behavior.

E2E validation completed successfully on `d7a5bdb587bbdad8f7ae29c40ad18fd73708156a`. Results: - `make test-e2e`: passed - Happy path: 10 passed, 0 failed - Platform/internal ports and encrypted routing: 8 passed, 0 failed - Environment parity: 9 passed, 0 failed - SSH-signed CLI operations: 9 passed, 0 failed - `make test-e2e-builder`: 10 passed, 0 failed - Dedicated builder completed, EIF and DB metadata were verified, the builder terminated, and the repeated build hit the cache. - BYOC E2E: 9 passed, 0 failed - `caution init --byoc`, customer-bucket builder deployment, attestation verification, and teardown all passed. The `make test-e2e-byoc` wrapper stalled while waiting on its `flock` despite no visible lock holder, so I ran the target body directly with the same configured environment, setup, published provisioner, test script, cleanup trap, and teardown. The underlying BYOC test passed cleanly. The happy-path suite reported its existing non-failing `No HTTP Gateway Configured` response warning. The same warning is present in earlier pre-branch E2E logs, so it does not appear to be introduced by this PR. All provisioned test resources and local test containers were cleaned up. No regression was found in current-image builds, dedicated-builder builds, BYOC builds, deployment, attestation, environment handling, or platform-port behavior.

No reviewers

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

public/platform!372

No description provided.

Rows
Columns