seems that saving a post triggers "create missing translations" task #59

Closed
opened 2026-07-20 18:42:46 +00:00 by hugo · 1 comment
Owner

when I save a post, I get a task named "Fehlende Übersetzungen werden erstellt" that I didn't expect. The first manual save must create the defined translations, later manual saves should not trigger any translations at all.

when I save a post, I get a task named "Fehlende Übersetzungen werden erstellt" that I didn't expect. The first manual save must create the defined translations, later manual saves should not trigger any translations at all.
hugo added the bug label 2026-07-20 18:42:46 +00:00
Author
Owner

Implemented in d1829f0.

Manual canonical post saves now calculate the currently missing configured languages before touching the AI endpoint or credential store and enqueue exactly one AI task per missing language. Saves of translation drafts, tab-switch/internal persistence, and editor link/media insertion do not schedule translations. Each worker recomputes the missing-language set, so a translation created after scheduling turns that worker into a safe no-op. Publish retains the specified explicit-trigger behaviour.

Cargo test binaries now use a process-local credential store, preventing test runs from opening a macOS keychain password prompt; normal RuDS binaries continue to use the operating-system keychain. The stale autosave statement in action_patterns.allium was aligned with editor_post.allium, translation.allium, and bDS2.

Red/green coverage verifies first-save task cardinality, later-save no-op, translation-draft no-op, internal editor persistence no-op, worker revalidation, and credential operations without the keychain. Verified with allium check/analyse, cargo test --workspace, cargo build --workspace, clippy with warnings denied, formatting, and diff checks. Final neutral review found no issue-specific gaps against #59, bDS2, or the Allium contract.

Implemented in d1829f0. Manual canonical post saves now calculate the currently missing configured languages before touching the AI endpoint or credential store and enqueue exactly one AI task per missing language. Saves of translation drafts, tab-switch/internal persistence, and editor link/media insertion do not schedule translations. Each worker recomputes the missing-language set, so a translation created after scheduling turns that worker into a safe no-op. Publish retains the specified explicit-trigger behaviour. Cargo test binaries now use a process-local credential store, preventing test runs from opening a macOS keychain password prompt; normal RuDS binaries continue to use the operating-system keychain. The stale autosave statement in action_patterns.allium was aligned with editor_post.allium, translation.allium, and bDS2. Red/green coverage verifies first-save task cardinality, later-save no-op, translation-draft no-op, internal editor persistence no-op, worker revalidation, and credential operations without the keychain. Verified with allium check/analyse, cargo test --workspace, cargo build --workspace, clippy with warnings denied, formatting, and diff checks. Final neutral review found no issue-specific gaps against #59, bDS2, or the Allium contract.
hugo closed this issue 2026-07-20 20:21:46 +00:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: hugo/RuDS#59