clanker/clanker-saddle @ 181fbd9926d10eb92e04eee7f1191e3120a7ef51

clanker 181fbd9926 r3: doc residue — the PendingAnswer docblock still linked the deleted SaddlePollEndpoint; it points at the SaddlePollState ledger now. (The goal page's own is_plan_review flag is GoalDetailEntrypoint's GoalBlockView member — live, not the deleted task-page BlockView twin.)		1 mese fa
..
Api	2cf9149d29 r3: legacy wire removal (arch 19, P3-1) — SaddleUpdateEndpoint + SaddlePollEndpoint and their routes are gone (the daemon never called them; verified against saddle/src at review time): the poll-era ingest/verdict pair 404s now, pinned by regressions. The service support that existed only for them dies with them: take_pending_answers (the saddle-keyed variant; the for_task assignment carrier stays), take_pending_comments + PendingComment, validate_token (its only caller was the update/poll bearer pin — resolve_bearer collapses to saddle_for_token for the all-id-less wire), the subtask/plan refusal arms, and the update endpoints' own statics move to SaddleWire (read_json_body/json_ok/json_error shared by every remaining /api/saddle/* route). Knock-ons per the plan: tasks.working_branch was write-dead (only /update done ever supplied it) — column + entity field + TaskDetailPublic member + the task-page line + mark_terminal/complete_task's dead parameter die; the plain-task plan-review UI in task_detail.html was unreachable (plans are goal-only since plan 12; only /update block type=plan could raise one) — BlockView.is_plan_review, the plan-review section, and the task page's ~180-line plan-remarks JS (dead with no [data-plan-review] targets) are removed; AnswerBlockAction's plan-verdict arms stay (the deep-link API + off-contract pushes keep working). Test fixtures rewrite into the push-era shapes: block raises ride POST /results blocked pushes (pages/lifecycle/push bundles), the terminal-task review fixture inserts its row directly (terminal pushes are consumed silently), rejected entries ride a results push, the dashboard review raise rides a review entry, the queued-state guard becomes a results state no-op, the /update subtask/plan refusals become 404 pins, and branch_advanced — the /poll verdict hint — is re-pinned on its surviving surface: the answers bundle's parked-resume allocation now asserts branch_advanced=true after a strictly-newer sibling completion on the shared branch. 856 -> 850 checks.	1 mese fa
Data	181fbd9926 r3: doc residue — the PendingAnswer docblock still linked the deleted SaddlePollEndpoint; it points at the SaddlePollState ledger now. (The goal page's own is_plan_review flag is GoalDetailEntrypoint's GoalBlockView member — live, not the deleted task-page BlockView twin.)	1 mese fa
Push	9f5ab7ab36 push: plan 11 task E — real Web Push (VAPID) notifications	1 mese fa
Services	13d283c0a6 r3: query layer (arch 14) — the work moves into the ORM with identical results. list_tasks filters (status/project) and sorts (priority, created_at — DateTime columns are unix-epoch ints, so the SQL ORDER BY equals the old .to_unix() compare) inside the query; get_ready_tasks stops loading EVERY task row for predecessor checks and queries only the queued candidates' predecessors; unreachable_task_ids batches (candidate tasks once, their sessions in one id-filtered query, the saddles in another — the newest-open-session-per-task semantics preserved by an id-ordered backward walk); display_map_for loads the wanted tasks' blocks in ONE task_id-filtered query instead of three full-table scans filtered in Vala; TasksState.build materialises the task list ONCE (the triple list_tasks died); TaskDetailState's enqueue dropdown queries exactly the standalone rows it lists (new list_standalone_tasks) instead of every task on every pulse; TimelineService.recent_window stops loading every session twice — the yeehaw session->task map builds from a batched lookup of only the WINDOW newest entries' sessions. The batched WHERE-in is an OR-chain helper (id_in/id_filter) because the expression grammar has no IN operator; id batches carry decimal strings (the Invercargill collections mishandle boxed value types — an int64 vector through to_immutable_buffer corrupts the heap, the exact trap the CreateTaskAction comment warns about). 851 checks, unchanged.	1 mese fa
actions	bf9009c91b r3: template partials (arch 16) — Statum DOES support includes (the mkpstm fragment pass: pstm-fragment with pstm-input parameters and pstm-materialise-as tag carry), so the partial strategy is compile-time fragments, not JS extraction. Five new partials in src/pages (auto-discovered, wired into template_deps): goal-branches (the branch chips paragraph, included 3x from goal_detail — head passes its note via the optional input, the note span hides where undefined), goal-compare (the Compare changes link, 2x), progress-feed (the latest+history card body shared between task and goal pages, parameterised on entries — heading and empty-state note stay per page with their page-specific wording), block-answer (the option-card question form — radio options + Other, shared between both pages' block loops; both loops expose the block as 'block' so no inputs), and block-reply (the free-text RFC reply form, shared likewise). The goal page's select+textarea answer form unifies onto the task page's richer option-card shape (the U1-pinned one): AnswerGoalBlockAction gains the __other__ resolution the task action already had, so a picked radio posts option=<value> exactly like the old select and the Other row carries the free text — the goal questions stay answerable with and without options. The goal page also gains the RFC reply fallback the task page already had (its old single form covered rfc blocks implicitly). Compiled-page inspection pins the composition: chips wrappers carry stm-if + the note def, no pstm-* residue. 851 checks.	1 mese fa
entrypoints	13d283c0a6 r3: query layer (arch 14) — the work moves into the ORM with identical results. list_tasks filters (status/project) and sorts (priority, created_at — DateTime columns are unix-epoch ints, so the SQL ORDER BY equals the old .to_unix() compare) inside the query; get_ready_tasks stops loading EVERY task row for predecessor checks and queries only the queued candidates' predecessors; unreachable_task_ids batches (candidate tasks once, their sessions in one id-filtered query, the saddles in another — the newest-open-session-per-task semantics preserved by an id-ordered backward walk); display_map_for loads the wanted tasks' blocks in ONE task_id-filtered query instead of three full-table scans filtered in Vala; TasksState.build materialises the task list ONCE (the triple list_tasks died); TaskDetailState's enqueue dropdown queries exactly the standalone rows it lists (new list_standalone_tasks) instead of every task on every pulse; TimelineService.recent_window stops loading every session twice — the yeehaw session->task map builds from a batched lookup of only the WINDOW newest entries' sessions. The batched WHERE-in is an OR-chain helper (id_in/id_filter) because the expression grammar has no IN operator; id batches carry decimal strings (the Invercargill collections mishandle boxed value types — an int64 vector through to_immutable_buffer corrupts the heap, the exact trap the CreateTaskAction comment warns about). 851 checks, unchanged.	1 mese fa
pages	bf9009c91b r3: template partials (arch 16) — Statum DOES support includes (the mkpstm fragment pass: pstm-fragment with pstm-input parameters and pstm-materialise-as tag carry), so the partial strategy is compile-time fragments, not JS extraction. Five new partials in src/pages (auto-discovered, wired into template_deps): goal-branches (the branch chips paragraph, included 3x from goal_detail — head passes its note via the optional input, the note span hides where undefined), goal-compare (the Compare changes link, 2x), progress-feed (the latest+history card body shared between task and goal pages, parameterised on entries — heading and empty-state note stay per page with their page-specific wording), block-answer (the option-card question form — radio options + Other, shared between both pages' block loops; both loops expose the block as 'block' so no inputs), and block-reply (the free-text RFC reply form, shared likewise). The goal page's select+textarea answer form unifies onto the task page's richer option-card shape (the U1-pinned one): AnswerGoalBlockAction gains the __other__ resolution the task action already had, so a picked radio posts option=<value> exactly like the old select and the Other row carries the free text — the goal questions stay answerable with and without options. The goal page also gains the RFC reply fallback the task page already had (its old single form covered rfc blocks implicitly). Compiled-page inspection pins the composition: chips wrappers carry stm-if + the note def, no pstm-* residue. 851 checks.	1 mese fa
Format.vala	375d0957f9 r1: one time-format util — src/Format.vala's format_stamp owns the '%Y-%m-%d %H:%M' display shape; the four per-entrypoint format_stamp/format_time copies and the saddles/rules/projects inline formats collapse onto it (StatePulse keeps its second-precision pulse stamp); the clock domain is uniform now_local (PushService's VAPID exp was the lone now_utc, and to_unix is timezone-independent anyway).	1 mese fa
Json.vala	34f58415c3 r1: one JSON-entry reading surface — src/Json.vala grows EntryJson (string_member/optional_string_member/get_string_or/string_array_member/parse_iso/parse_string_array); SaddleService's four private statics, SaddleApi's get_string_or/parse_iso/string_options and TaskService.parse_block_options all collapse into it, three string-array parsers down to one, behaviour byte-identical (the malformed-options warning keeps each call site's wording).	1 mese fa
main.vala	e236aa7079 r3: api_tokens dropped (arch 24, P3-2) — saddles.token is the single credential store: the legacy table, its entity mapping, the startup ensure_saddle_tokens migration (main.vala's call + the service), token_for_saddle (the saddles page and the /start callback token read the row directly), unregister's sweep, and saddle_for_token's fallback read all die. Nothing deployed predates the plan-10 row token (operator decision), so no credential loses validity — a legacy-shaped token string now 401s everywhere, pinned by the rewritten restart scenario: the fresh schema carries no api_tokens table, the row token survives a restart unchanged and still authenticates, and the legacy-shaped string authorises nothing. 850 -> 851 checks.	1 mese fa