Remove post_import from the docs (it never actually runs)#884
Conversation
post_import is a documented, parsed, and validated app.toml key that the runtime never actually executes, so the docs were pointing users at a step that silently no-ops. Its headline use case is database migrations, which makes the trap especially sharp: the deploy reports success while the schema never changes, and the app breaks at runtime. Pull it from the app.toml reference and the terminology glossary so it stops trapping people. The larger wire-vs-retire call (and the code-level handling of the key) follows separately under MIR-1289.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (2)
💤 Files with no reviewable changes (2)
📝 WalkthroughWalkthroughThis pull request removes documentation references to the Changes
Related PRs: None identified. Suggested labels: documentation Suggested reviewers: None identified. Poem: Comment |
phinze
deleted the
phinze/mir-1289-post_import-in-apptoml-is-parsed-and-documented-but-never
branch
2 participants
post_importlooks like a real feature from the docs: it's a documentedapp.tomlkey with precise semantics ("runs after a new version is imported but before it receives traffic, commonly used for database migrations"), and it's parsed and validated on the way in. The catch is that nothing in the runtime ever reads it. The Go field is referenced exactly once in the whole repo, at its own declaration. So a deploy withpost_import = "…/rake db:migrate"reports success while the migration never runs, and you find out later when a worker crash-loops on a missing table.That's the worst kind of trap because the docs actively lead you into it. This pulls
post_importout of the app.toml reference and the terminology glossary so it stops advertising a step that doesn't happen. Purely docs here, no behavior change.The bigger "wire it or retire it" decision has landed on retire, but the code-level piece (warning on the key instead of silently accepting it) and folding the escape-hatch capability into MIR-853's one-shot service work comes in a follow-up. This is just the urgent de-trapping.
Part of MIR-1289.