@m-kopa/launchpad-cli 0.71.0 → 0.73.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -6,6 +6,21 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
6
6
  This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html);
7
7
  pre-1.0 minor bumps may carry breaking changes per ADR 0005.
8
8
 
9
+ ## 0.73.0 - 2026-10-06
10
+
11
+ ### Fixed
12
+
13
+ - One `launchpad deploy` now ships a **new container app** whose `launchpad.yaml` declares `env-file` secrets (AIO-892). Before, the first deploy always stopped at `deployment_verified` because the deploy workflow found the secrets missing, and recovering cost `secrets push --redeploy` plus `launchpad recover`, which spends the single same-tree retry. Now `deploy` checks `.env` before it calls the bot (a missing or empty declared name stops the deploy with the names listed and nothing sent), and once the bot has committed the app repo it pushes each declared secret with the same code as `launchpad secrets push` and keeps watching. The bot holds the deploy run's authorization for up to about 3.5 minutes until the secrets exist. `launchpad secrets push` retries a `manifest_not_found` reply for about 30 seconds while the app repo is being created. The two-command recovery still works as the fallback, and `launchpad recover` now refuses (409 `missing_container_secrets`, names only) before it spends its one retry if the declared secrets are still absent.
14
+ - `launchpad recover` for a container app now actually completes: its same-tree retry commit starts a deploy run that fails at the workflow's `authorize` job unless the bot publishes the exact-run control record, which nothing did. The bot now starts a Workflow instance for the retry that finds the run, publishes the record, verifies the deployment and marks the app `live`, so one `launchpad recover <slug>` is enough (a second call never makes a second retry commit). A retry only counts as used if the platform actually authorized its run: when the retry commit's run failed without any control record having been published (the old bug), `recover` re-runs that same run (no new commit) up to twice instead of answering `container_recovery_exhausted`. `launchpad recover` no longer prints a `Pages project ... MISSING` line for container apps, which have no Pages project.
15
+
16
+ ## 0.72.0 - 2026-10-06
17
+
18
+ ### Fixed
19
+
20
+ - `launchpad secrets push` now works for a container app that has not deployed yet. A container's deploy workflow reads its declared secrets from GitHub Actions secrets on the app repo, which authors cannot reach, so `push` also writes them there (ADR 0059): only names declared in `secrets.bindings[]`, never platform-owned names, encrypted to the repo's key. A Worker that does not exist yet is reported as `skipped_not_deployed` without failing the push. Before this, a container's first deploy failed at "Verify required deployment inputs" and every later deploy would have too.
21
+ - For a container app stuck at `provisioning_failed` / `deployment_verified` because of missing secrets, `push` prints the next step: `launchpad recover <slug>` (run it again when the deploy run finishes). `launchpad secrets push --redeploy` runs the first `recover` for you; it has no effect on an already-deployed container app, and `launchpad redeploy` stays a no-op for containers.
22
+ - `launchpad secrets status` lists a `github_actions` surface for container apps.
23
+
9
24
  ## 0.71.0 - 2026-10-06
10
25
 
11
26
  ### Added