feat(config): trigger the config.toml migration at startup#245
Merged
Conversation
ShouldMigrate fires when config.toml exists and state.toml doesn't: state.toml is only written on success, so its absence doubles as the "not migrated yet" marker and an aborted run retries on the next one. Migrate is a deliberate no-op until the migration body lands. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Up to standards ✅🟢 Issues
|
| Metric | Results |
|---|---|
| Complexity | 33 |
| Duplication | 3 |
TIP This summary will be updated as you push new changes.
…toml (#246) Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
f17369c
into
feat/profile-migration
1 of 2 checks passed
LorrisSaintGenez
added a commit
that referenced
this pull request
Jun 15, 2026
…#248) * feat(config): trigger the config.toml migration at startup (#245) Co-authored-by: Claude Fable 5 <noreply@anthropic.com> * chore(config): slim down the migration comments Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(config): harden the migration against bad config.toml files Two startup-path regressions found while testing edge cases: - An undecodable profile (root-level scalar, unconvertible field type) hit ConfiguredProfiles' log.Fatalf and bricked every command, --help included, forever (state.toml never written). The migration now decodes profiles itself and skips undecodable entries with a log. - An unparseable config.toml was silently read as zero profiles, so an empty state.toml marked the migration as done forever. Migrate now aborts when config.toml cannot be parsed and retries on the next run. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(config): migrate search_hosts and crawler_user_id to state.toml The remaining non-secret profile data moves with the migration so it survives the eventual config.toml removal. GetSearchHosts and GetCrawlerUserID gain a new-model branch: the resolved application's state.toml entry answers first, an empty value falls through to the legacy config.toml lookup while both models coexist. admin_api_key stays excluded as decided on the ticket. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * refactor(config): omit empty api_key_uuid from state.toml Legacy migrated keys have no UUID, so without omitempty every migrated entry serialized a noisy api_key_uuid = "". The field is only set by new-model writes (app create/select); omitting the empty value matches search_hosts/crawler_user_id and reads back identically. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(application): mark configured apps in select from state.toml The select picker decorated choices from config.toml profiles only, so apps configured under the new model showed as unconfigured. It now marks an app "(configured)" when state.toml holds an entry for it, falling back to legacy config.toml profiles while config.toml is still supported. Adds Config.ApplicationInState for the state-only lookup. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(application): state-aware "configured" marker + describe without auth Two follow-ups from auditing the command tree against state.toml: - application list marked apps configured from config.toml profiles only, so new-model apps showed as "(not configured)". It now uses the same state-first/config-fallback check as the select picker, centralized in apputil.ApplicationConfigured. - describe walks the command tree and needs no credentials, but lacked skipAuthCheck so it failed on a machine with nothing configured. It now skips the auth check. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(config): clearer error when the current app has no keychain key When state.toml resolves a current application but its key isn't in this machine's keychain (e.g. state.toml synced across machines without it), GetAPIKey/GetCrawlerAPIKey returned the generic "not configured yet". The error now names the application and points to the fix (`application select` / `auth crawler`, or the matching env var). Re-authenticating rewrites the keychain entry, so it restores a working state. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * perf(config): short-circuit ShouldMigrate on the state.toml check Check state.toml first: an already-migrated machine (the steady state, hit on every command) now settles in a single stat instead of also stat-ing config.toml. The boolean result is unchanged. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * perf(application): O(1) configured-app lookup when listing/selecting Marking apps "(configured)" called ConfiguredProfiles() (a full viper re-parse) once per application — O(apps × profiles) with a heavy constant. Now the config.toml profile app IDs are collected once into a set (ProfileApplicationIDs); the per-app check is two O(1) map lookups (cached state.toml + the set). Both `application list` and `application select` build the set once before their loop. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Step 1 of GROUT-363 — wires the one-time config.toml → state.toml + keychain migration trigger, without the migration body yet.
Config.ShouldMigrate(): true whenconfig.tomlexists andstate.tomldoesn't.state.tomlis only written when the migration (or any new-model write) succeeds, so its absence doubles as the "not migrated yet" marker — an aborted migration naturally retries on the next run, no separate flag needed.Config.Migrate(): deliberate no-op for now — it must NOT writestate.toml, otherwise the trigger would stop firing before the real migration ships. The body lands in the follow-up PRs (per-profile move, skip rules, atomicity).Execute(), right afterInitConfig(): the migration has to run before the command executes because credential resolution readsstate.tomlonce per invocation and caches it. A migration failure never blocks the command (logged in debug mode only).Test
Unit tests only —
Migrateis still a no-op, so there is no observable CLI behavior yet:go test ./pkg/config -run TestConfig_ShouldMigrate -vGROUT-363