mirror of
https://github.com/windmill-labs/windmill.git
synced 2026-10-21 00:03:31 +00:00
5b1fd00aab570cc756fc0f69b592b568bba5c9cb
374
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
  Ruben FiszelandClaude Opus 4.8
|
4cf44280ce |
Merge branch 'main' into feat/asset-graph-view
Resolve conflicts in scripts.rs (import superset; ScriptWDraft/get-draft endpoint removed by main's Db-backed drafts refactor), CreateActionsFlow (keep Pipeline option over main's skipModal), scripts/add stub (main's redirect-to-edit), and 8 trigger editors (combine fixedScriptPath with main's draft-overlay loadTrigger). Regenerated backend client. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
 
|
1fc355709c |
feat: Db-backed user drafts (#9351)
* Db draft removal * refactor: drop unsaved-changes confirmation modal from editors * fix: remove nodraft from flow row edit link * fix: remove nodraft from app and raw app edit buttons * fix: remove nodraft from all edit links * fix: merge backend defaults into legacy autosaves to avoid spurious restore toast on raw apps * feat: add username column to draft table for user-scoped drafts * feat: add sync_drafts and list_users_with_draft_on_path endpoints * feat: add UserDraftDbSyncer service for bi-directional draft sync * feat: wire UserDraft.save through DbSyncer + conflict modal * refactor: gate useLocalStorageValue nested-update effect behind opt-in flag * refactor: move sync force flag from request-level to per-entry * feat: sync all userdraft kinds, switch draft owner to email FK, add id PK, scope draft list to readable paths * refactor: route draft permission check through authed.folders + RLS, drop client-supplied email * feat: support draft deletion via sync (value: null) with same conflict semantics * feat: surface other users' drafts in editors with diff+fork action * refactor: unify draft schema migrations and type kinds via DRAFT_KIND enum * perf: add (workspace_id, email, created_at) partial index for sync hot path * chore: update ee-repo-ref to a30079e75dc5b7d7413aa8ee20e40e80bfea9cbd This commit updates the EE repository reference after PR #597 was merged in windmill-ee-private. Previous ee-repo-ref: 55c19293232be379a3044eb78f677b545882ffd6 New ee-repo-ref: a30079e75dc5b7d7413aa8ee20e40e80bfea9cbd Automated by sync-ee-ref workflow. * fix(userdraft): trigger sync on deep mutations via readFieldsRecursively * Rollback UserDraft * remove queuing logic * pushDrafts * refactor: remove draft sync layer and conflict modal * feat: add save_draft, list_drafts, get_draft routes * feat: add get_draft overlay to getScriptByPath * feat: extend get_draft overlay to flow, app, resource, variable, schedule, triggers * feat: support null value in save_draft for deletes * readLastSyncMap * feat: redirect /add pages to /edit/draft_uuid with new_draft flag * fix: inline get_draft query field instead of flattening * fix: drop dangling nobackenddraft assignment in flows edit * feat: include user drafts in list endpoints with is_draft flag * fix: prefix draft paths with u/{user} and seed editor state on new_draft * fix: route draft-only deletes through UserDraftDbSyncer on home page * feat: delete user drafts when their underlying item is deleted * fix: empty path seed on new_draft so friendly auto-name fires * feat: re-add Draft and Draft only badges on home page rows * fix: synthesize value wrapper on draft-only raw_app response * fix: tolerate missing latest-version on draft-only flow reload * fix: skip first observable change in DB sync effect to match LS persist * fix: remove URL-hash sync from script editor (already marked TEMP) * refactor: drop localStorage layer from UserDraft * refactor: drop vestigial LS-era code from UserDraft * feat: migrate localStorage drafts to DB on layout mount * fix: migrate session runtime + script view to per-user draft API * feat: add 'Reset to deployed' action on draft-loaded toast * feat: hide 'Reset to deployed' action when no deployed version exists * createCoalescingKeyedRunner * example ts doc * createDebouncerByKey * refactor: drop await on draft-delete in reset flows, refetch deployed directly * fix: bridge saved-draft shape to wire shape in apps/resources/variables loaders * feat: route UserDraftDbSyncer.save through debouncer + coalescing runner * feat: add immediate-save bypass that cancels pending debouncer + runner tasks * fix: seed UserDraft cell from spec defaultValue on acquire * fix: redirect /add routes at load phase to eliminate white flash * fix: drop +page.js files in /add routes that conflicted with +page.ts * refactor: send draft as separate .draft field instead of deep-merging onto deployed * feat: surface draft path in home list when user typed one different from URL * feat: add UserDraft.stopSync/restartSync, wire on script + low-code app /add init * fix: thread URL path into ScriptBuilder.stopSync (was using empty initialPath) * fix: also stopSync in route's new_draft branch + queue pre-acquire suspensions * feat: add AutosaveIndicator backed by reactive UserDraftDbSyncer.getState * refactor: drop draft-loaded toast in non-route editors, banner now compares draft vs deployed * fix: gate per-user draft-only rows in listings on include_draft_only flag * feat: flush pending draft saves via keepalive fetch on tab hide / pagehide * autosave indicator nits * fix: route create-vs-update on /add deploys; seed policy.execution_mode; sync script template * chore: add [draft-sync] console logs to trace script bootstrap autosave * fix: seed auto-generated path in script new-draft route to suppress Path widget's autosave-triggering mutation * fix: defer script restartSync until script.path lands (Path widget gated on $userStore + $workspaceStore) * fix: poll script.path via tick() until Path widget settles before restartSync * chore: log inferArgs underlying error on deploy to diagnose 'Could not parse code' toast * fix: wait for script.path to stabilize across two ticks before restartSync * revert: drop unsuccessful path-stabilization heuristics + leftover [draft-sync] logs * fix: seed new-draft script schema as emptySchema() so inferArgs doesn't trip on undefined properties * fix: heal legacy drafts with schema={} (no .properties) on deploy * autosave indicator * refactor(editors): drop UnsavedConfirmationModal mount + Show diff button * feat(drafts): collaboration banner, cross-tab conflict detection, raw app template picker - Other-users-drafts banner (Modal2): the deployed-overlay response now carries `other_drafts_users` (workspace usernames only, never emails); each row offers View JSON + Fork. Drops the standalone `listUsersWithDraftOnPath` endpoint; `getDraftForUser` now takes a workspace `username` query param (resolved to email server-side). - Cross-tab/browser save conflict detection: the syncer attaches `last_sync` to every save (defaults to non-force); on a `conflict` response it parks a snapshot in a reactive map. Each route mounts a `DraftSyncConflictModal` and seeds the per-tab `last_sync` via `recordRemoteSync(query, draft_saved_at)` on every `get_draft` load. Keepalive flush also respects optimistic concurrency. - Raw app template picker re-added after the /add ⇒ /edit refactor: framework (React 19 / 18 / Svelte 5), data table + schema config, and optional AI prompt — extracted into `RawAppTemplatePicker.svelte` and driven by `new_draft=true` on the edit route. * fix(drafts): suppress autosave during /add template seeding on script + raw app editors - ScriptBuilder: delay `restartSync` 500ms past `initContent` + stores- ready so the Path widget's `$workspaceStore && $userStore`-gated `initPath → reset → onMetaChange → bind:path` cascade lands inside the suspension window. Two `tick()` waits weren't enough — the bind:path mutation fired ~100ms after the prior `restartSync` and posted as a "user edit". - apps_raw route: suspend autosave on `new_draft=true` and resume only after the framework picker closes (via `onStart` or X dismissal), with a two-tick settle so the picker's seeded `files/runnables/data/policy` mirror to `draftHandle.draft` observably advances `lastSerialized` before sync re-arms. * fix(drafts): land /add redirects on the real workspace username, not "me" The `/add` → `/edit/u/{username}/draft_{uuid}` redirects ran during SvelteKit's load phase, BEFORE the (logged) layout's async `getUserExt` populated `userStore`. `get(userStore)?.username` returned undefined and fell back to the `'me'` placeholder on every fresh nav, producing `u/me/draft_{uuid}` paths instead of the user's real namespace — broke ownership checks against `authed.username` and silently scoped autosaves under the wrong path. Layout now persists `username` to localStorage on every successful `getUserExt`, and `getUsernameForNamespace` (new shared helper, used by all four `/add/+page.ts` files) reads the live store first, falls back to the cached value, and only then to `'me'` for true first-ever loads. * fix(drafts): key low-code app autosave on the URL path, not the empty string `AppEditor` keyed its `UserDraft.use` handle on `newApp ? '' : path` — a legacy leftover from when `/apps/add` was its own URL (no path). With the `/add` ⇒ `/edit/u/{user}/draft_{uuid}` redirect, `newApp=true` made autosaves land on the `('app', '')` row instead of the URL path: - The `apps/list?include_draft_only=true` query joins drafts onto `app.path`, surfacing drafts at the URL path. The empty-path row didn't match the user's URL so the draft never appeared in the home list. - Refreshing `/apps/edit/u/{user}/draft_{uuid}` re-fetches at the URL path with `?get_draft=true`, finds nothing, and 404s. Drop the ternary so the handle always uses `path` — the same as scripts/flows/raw_apps. The route's `?new_draft=true` branch already seeds the empty-template baseline, so there's no longer a "the draft sits under '' until first save" race to worry about. * fix(raw_app): propagate template picker X / Esc dismissal so autosave resumes The picker mounted `<Modal kind="X" open ...>` (one-way prop, not `bind:open`). When the user dismissed via X / Esc / click-outside, the inner Modal flipped its own local `open` to false (hiding the UI) but never wrote back to the picker's `open` $bindable. The route's `templatePicker → false` watcher — the one that calls `restartSync` two ticks after the picker closes — never fired, so autosave stayed suspended and the user's edits after dismissal were silently dropped. Switch the inner Modal to `bind:open` so the dismissal bubbles all the way up to the route's state. "Start without AI" already worked because its `onStart` handler explicitly sets the picker's `open = false`. * nit unused * fix(drafts): make the home-page View/Edit JSON action work on draft-only apps The "View/Edit JSON" entry on the home page called `AppService.getAppByPath` without `get_draft=true`, so for draft-only items at `u/{user}/draft_{uuid}` the backend 404'd with "App not found at path …". Pass `get_draft=true` and render the synthesized stand-in's editable shape: - App drafts come back as `{summary, value, path, policy, ...}` — `value` is the App definition the editor was working on; show that. - Raw-app drafts come back as the flattened `{files, runnables, data, summary, policy, ...}` with no nested `value`; show the whole shape. On save, draft-only items can't go through `updateApp` (no deployed row). Route the edit through `UserDraftDbSyncer.save` (with `immediate: true` so `await` resolves after the POST lands) and relabel the button "Save draft" + Save icon. Deployed items keep the existing "Deploy" flow unchanged. * fix(drafts): render the right shape in View/Edit JSON for draft-only items The previous fix landed `fapp.value` into the editor, but the deployed-overlay flattens the bare editable shape into `inner`/the top-level response — drafts have no nested `.value`. So: - App drafts (`{grid, breakpoints, hiddenInlineScripts, …}`) rendered as empty (`fapp.value` was undefined). - Raw-app drafts 404'd outright: `get_draft=true` with no `rawApp` flag can't tell which draft kind to look up, defaults to `app`, doesn't find one. Thread the row's `raw_app` flag from AppRow → `appExport.open(path, rawApp)` → `getAppByPath({..., rawApp})` so raw-app drafts resolve to the right `UserDraftItemKind`. Read `fapp.draft` (the bare editable shape from `fetch_draft_only`) into the JSON editor for draft-only items — clean payload, no `is_draft` / `no_deployed` / overlay noise. Save the same bare shape back through the syncer so the regular editor reads it unchanged on the next mount. * fix(drafts): skip public-secret-URL fetch in the Deploy drawer for draft-only apps Opening the Deploy drawer on a `/edit/u/{user}/draft_{uuid}` app fired `AppService.getPublicSecretOfApp` immediately because the gating effect only checked `appPath != ''` + `savedApp`. The `/secret_of/{path}` route plain-SELECTs `app.id`, so a draft-only path 404'd with "App not found at name …" and the public-URL ClipboardPanel spun forever waiting on `secretUrl`. Thread the existing `newApp` signal (already on `AppEditorHeader` / `RawAppEditorHeader`) into `AppEditorHeaderDeploy`, gate the fetch behind `!newApp`, and render the existing "Deploy this app once to get the public secret URL" placeholder instead of the spinner for draft-only items. * fix(drafts): disable Diff button on draft-only items across the 4 editors Diff has no baseline to compare against on draft-only items — the button used to be gated by the pre-PR `/add` route's own state, but the `/add → /edit` redirect landed everything under the regular `/edit` page where the gate was missing. - ScriptBuilder: gate the topbar Diff on `savedScript.no_deployed`; seed `no_deployed: true` on the route's `new_draft` empty NewScript so the gate fires before the first deploy. - FlowBuilder: gate the topbar Diff on `newFlow` (route already sets it from `backendFlow.no_deployed` and the new-draft branch). - AppEditorHeader: gate both the "Diff" dropdown action and the Deploy-drawer's "Diff" button on `newApp`. - RawAppEditorHeader: gate the topbar Diff + the Deploy-drawer's "Diff" button on `newApp`. Each gate also rewrites the tooltip ("Deploy this … once to compare against the deployed version") so the hover state explains why. * fix(drafts): disable the "No login required" toggle on draft-only apps Flipping the toggle called `setPublishState`, which POSTs the new `policy` through `AppService.updateApp` — that handler's `UPDATE app ... RETURNING path` finds nothing on a draft-only path and `not_found_if_none` 404s with "App not found at name …" (apps.rs:1975). Gate the Toggle on `!newApp` too so the user has to deploy once before configuring the publish state. * refactor(drafts): drop dead draft_path field from list responses The draft-only listing branches in scripts/flows/apps computed a `draft_path` from the draft JSON (when the user-typed path differed from the URL's autogenerated `u/{user}/draft_{uuid}`), and `{Script,Flow,App} Row.svelte` preferred it over `path` for the row title. In practice that path is never written: the app, raw-app and flow editors all warn "Deploy the X to make the path change effective" — the rename only lands on deploy, never in the draft. So the field is always None and the home rows always show the autogenerated slot anyway. Drop the field from the three `Listable*` structs, the three draft-only push sites, the three OpenAPI response schemas, and the three frontend row components. Client regenerated. * fix(drafts): seed a friendly name on /flows/add The flow route passed `initialPath={page.params.path ?? ''}` to FlowBuilder, so on the `/flows/add → /flows/edit/u/{user}/draft_{uuid}` redirect the Path widget's `initPath` saw a non-empty `initialPath` and skipped the `reset()` branch that auto-generates the friendly `<random_adj>_flow` name. The other three editors all clear `initialPath` in their `new_draft` branch for exactly this reason. Track `initialPath` as route-owned state (defaults to the URL path) and clear it to '' inside the `new_draft` branch, then bind it through to FlowBuilder so any post-deploy update from the editor still propagates. * feat(drafts): render friendly user-typed path on home list for all 4 kinds Reinstate `draft_path` on `Listable{Script,Flow,App}` so the home rows prefer the user-typed name over the autogenerated `u/{user}/draft_{uuid}` URL slot, with two source rules — one per how each editor wires the Path widget: - Scripts already work: `ScriptBuilder` binds the Path widget directly to `script.path`, so the typed path round-trips through the draft JSON's own `path` field. Backend extracts `v["path"]` when it differs from `row.path`. - Flows / apps / raw apps don't write the typed path into the autosaved value (`Flow.path` is one-way-bound to `$pathStore`; the bare `App` / raw-app value has no `path` field at all). Introduce an explicit `draft_path` field on the draft JSON, written by the editor ONLY when the typed path differs from the deployed/seeded `savedX.path`: - FlowBuilder: $effect on `$pathStore` mutates `flow.draft_path`. - AppEditorHeader: $effect on `newEditedPath` mutates `$app.draft_path`. - RawAppEditorHeader: $effect surfaces `pendingDraftPath` up via the bind chain (RawAppEditor → route); the route's draftHandle.draft spread includes `draft_path` when set. Backend extracts `v["draft_path"]` and `None` when unchanged or after deploy (deploy clears the whole draft, so the field naturally disappears post-deploy without bookkeeping). Flow route's `new_draft` branch now stops sync around the Path widget cascade, with a 700ms scheduled `restartSync` (mirrors the existing scripts/apps/raw_apps stoppers) — the new draft_path mutation lands inside that window so `/flows/add` no longer fires an autosave before the user's first edit. openapi/sqlx regenerated. * fix(drafts): preserve the user-typed draft_path on reload of draft-only items The flow / app / raw-app editors all dropped the saved `draft_path` back to the URL's `u/{user}/draft_{uuid}` slot the moment the user reloaded a draft-only edit page: the route sourced the Path widget's initial path from `page.params.path` instead of the previously-saved `draft_path`, and the first user edit then mirrored that URL path back into the autosaved draft — silently overwriting the friendly name in both the row and the editor. - Flow route: after computing `effectiveFlow`, override `flowInitialPath` with `effectiveFlow.draft_path` when set. - App route: pass `newPath={(app.value as any)?.draft_path ?? app.path}` through to `AppEditor`; AppEditorHeader's `newEditedPath` default now prefers a non-empty `newPath` over the random `<adj>_app` seed (the `newApp && !newPath` branch keeps the `/apps/add` friendly auto-name). - Raw-app route: surface `savedRawAppDraft.draft_path` onto `backendApp` so the `extractRawApp` path seeds `newPath` with the friendly name. Reload + a subsequent edit now leaves `draft_path` intact for all three kinds; verified end-to-end via the `/drafts/get_draft/...` endpoint. * fix(ui): default Modal2 target to 'body' so omitting the prop doesn't throw Modal2 defaulted `target = ''` and forwarded it to `Portal`, which calls `document.querySelector(target)` — an empty selector throws "Failed to execute 'querySelector' on 'Document': The provided selector is empty" and the modal silently fails to mount. That's why `OtherUsersDraftsModal` (and `DraftSyncConflictModal`) never appeared on editors where another user had a draft — both omit the `target` prop. Other Modal2 callers (StorageSettings, CriticalAlert, CustomInstanceDbWizardModal, …) pass an explicit `target="#content"` and were unaffected. Match Portal's own default of `'body'` so omitting the prop is now a no-op rather than a runtime throw. * fix(drafts): Reset to deployed no longer resurrects the draft The toast's "Reset to deployed" callback POSTed `value: null` to the syncer, then handed control to the route's `onResetToDeployed` (which wipes the in-memory handle and reloads the deployed payload via `getDraft: false`). Both writes flowed through the reactive sync effect: the wipe scheduled a delete, the reload scheduled a re-save of the deployed value as the new draft. Coalescing collapsed them and the draft came back — making the "discard" action effectively a no-op. Wrap the whole callback in `UserDraft.stopSync` / `restartSync`. The explicit `value: null` POST still goes through (it's a direct `UserDraftDbSyncer.save` that doesn't depend on the reactive effect), the route's wipe-then-reload mutations advance `lastSerialized` silently under suspension, and the next user edit (after two ticks past the deployed-seed write) is the first real save again. * ui nit * feat(drafts): autosave-indicator popover with Reset-to-deployed action Click the cloud icon → popover with "All changes are saved as a draft on the server. The draft is per-user — your teammates' editors keep their own." When the editor isn't on a draft-only path AND the user has a draft (UserDraft.has returns true), a "Reset to deployed" button mirrors the load-time toast action — stops sync, POSTs `value: null`, runs the route's reload-without-draft callback, restarts sync past two ticks so the deployed-seed write doesn't resurrect the draft. Threaded `onResetToDeployed` from each route down to its builder (ScriptBuilder / FlowBuilder / AppEditorHeader / RawAppEditorHeader) and into the indicator. `draftOnly` is wired from `savedScript.no_deployed` / `newFlow` / `newApp` so the action hides where there's nothing to fall back to. The indicator's trigger now has a hover affordance + matches Portal's default target ('body') via Modal2's earlier fix. * fix(drafts): wait for the fork POST to land before navigating OtherUsersDraftsModal's Fork action called UserDraft.save, which routes through the autosave debouncer (1500ms). The subsequent goto fired within the same tick, so the destination editor's get_draft=true read ran before the POST landed and 404'd — refreshing worked because by then the debounced save had fired. Call UserDraftDbSyncer.save with immediate: true and await it. The syncer cancels any queued debouncer task for the key and resolves the promise only after the POST completes, so the route load can find the forked draft on the first try. * fix(drafts): conflict detection — keep last_sync map tab-local instead of in localStorage Two tabs editing the same draft both load with last_sync = T0. Tab-1 saves; the server accepts, returns T1, and the syncer wrote T1 into localStorage. Tab-2 then tries to save: it reads the SHARED localStorage map, sees T1 instead of its own baseline T0, sends last_sync = T1, and the backend's WHERE clause (`created_at <= last_sync`) is true → tab-2 clobbers tab-1's edit without ever seeing a conflict. Move the map to tab-local memory (`new Map<string, …>`). Reload of the tab now starts with an empty map; that's fine because the editor's load path calls `recordRemoteSync(query, draft_saved_at)` right after `get_draft=true` returns, reseeding from the authoritative server timestamp before any user edit could fire a save. * fix(drafts): OtherUsersDraftsModal — close on Fork, don't leak clicks through nested JSON Two bugs in the per-editor "another user has a draft" banner: - Fork landed the immediate save but didn't close the banner before navigating. Svelte hadn't torn down the previous route's components by the time goto returned, so the banner lingered on top of the destination editor. Comment the explicit isOpen=false on the happy path so it's clear it MUST run before goto. - Clicking anywhere on the screen while the View JSON drilldown was open closed the underlying banner too. Modal2's clickOutside action fired on every Modal2 instance — both the JSON modal and the underlying banner — because both attach their own listener at the document level. Add `closeOnOutsideClick` opt-out on Modal2 and pass `closeOnOutsideClick={!jsonOpen}` to the outer modal so clicks outside the JSON drilldown only close the drilldown. Drive-by: Modal2's keydown handler now ignores Escape when its own isOpen is false (was a no-op closer that would still preventDefault on every key press, swallowing key events for any siblings). * fix(drafts): conflict modal wording — drafts are user-scoped, not teammate-scoped * fix(drafts): defer reset-to-deployed restart until first user interaction Two-tick `restartSync` was too aggressive: editor remounts emit a tail of cascading writes (Monaco setValue acks, schema re-infer, UI Builder iframe handshakes, schedule-config recomputes, …) that land well after two ticks and would clobber the just-deleted draft with an upsert of the deployed value — making "Reset to deployed" a no-op in practice, the user kept seeing the draft come back. Centralise the suspension lifecycle in a new `runResetToDeployed` helper. It stopSyncs around the reset, POSTs the explicit delete, runs the route's wipe-and-reload, and then arms a one-shot listener on document keydown / input / pointerdown that restartSyncs on the user's next real interaction. A 5-second fallback re-arms sync if the user walks away without touching the editor, so suspensions don't leak. Use it from both the load-time toast (`notifyDraftLoaded`) and the autosave-indicator popover so the two stay in sync — fixes both entry points. * indicator ui nits * fix(drafts): split tab-switch and unload flushes — kill self-conflict on visibility change The single keepalive flush bound to both `visibilitychange → hidden` and `pagehide` self-conflicted on tab switch: visibilitychange fires on every tab/app switch with the page still alive, the keepalive POST advanced the server's `created_at` to a fresh `now()`, the client discarded the response (no listener), the local `lastSync` stayed at the old value, and the next foreground autosave sent that stale timestamp → server saw `created_at > last_sync` → conflict modal for the user's own background-tab write. A still-pending debouncer task made it worse: it fired a second runner POST after the keepalive with the same stale `last_sync`, the second self-conflicted too. Split into two paths: - `visibilitychange → hidden` → `flushOnVisibilityHidden`: route through the normal runner pipeline. The page is alive, so the response can land and `setLastSync` keeps the baseline current. Call `debouncer.cancel(key)` first so a queued keystroke can't double-fire with the same stale `last_sync`. - `pagehide` → `flushOnPageHide`: keep the `keepalive: true` raw fetch for the genuinely-going-away case (the JS context is torn down, the response is necessarily discarded). Same `debouncer.cancel(key)` guard. On the next mount, the route's `recordRemoteSync(query, draft_saved_at)` reseeds `lastSync` from authoritative server state before any user edit can fire a save. * fix(drafts): drop the visibilitychange flush — debouncer keeps running on hidden tabs Tab switching just hides the page; the JS context survives and the debouncer's `setTimeout` keeps counting down. When it fires, the runner POSTs normally and the server's response updates `lastSync`. There's nothing left for a visibilitychange-driven flush to do that the ordinary pipeline doesn't already handle, and adding one only creates extra POSTs to reason about. `pagehide` remains the single trigger for the keepalive flush — that's the case where the JS context is actually being torn down and the runner's pending fetch would otherwise be killed mid-flight. * nit * refactor(drafts): drop LS-era pipeline; backend is canonical on load The PR's iteration left behind a meta/staleness pipeline carried over from the localStorage era — per-rev tracking, a LocalDraftStaleModal, a 'Restored from local storage' toast, and a localDraft-vs-backend comparison branch in every editor loader. With drafts now living in the DB and the optimistic-concurrency lastSync check handling divergence, that whole stack is dead weight. Worse, the comparison branch caused 'Load from server' in the conflict modal to do nothing: the loader preferred the in-memory cell over the backend, so the user-clicked 'load from server' just re-displayed the local edits AND fired two confusing toasts (Restored from local storage + Loaded your saved draft). The rip: * userDraft.svelte.ts: drop UserDraftMeta, StoredDraft.meta, checkStaleness, UserDraftStalenessCause, normalizeForCompare, localDraftDiffers, saveMeta, getMeta, setDraftAndMeta, setMeta, handle.meta/setDraftAndMeta/setMeta, force option. Handle is now just { draft }. * userDraftToast.ts: drop notifyRestoredFromLocal + RestoreFromLocalActions. Update copy. * LocalDraftStaleModal.svelte: deleted. * AppEditor.svelte: drop initialRevs prop and the firstMirror wipe-then-restore dance (it existed only to consume the meta-mismatch skip slot). * All 4 editor routes: backend is canonical on load — the in-memory cell is overwritten with the deployed+draft overlay, the syncer's seed guard swallows the first write so we don't POST it back. * VariableEditor / ResourceEditor: drop the staleness pipeline + rev bookkeeping; backend wins on open. * useTriggerDraftSync.svelte.ts: inline the JSON-normalize + deepEqual utility as a private cfgDiffers helper (kept for the form-vs-deployed dirty check, which is a genuine semantic compare, not LS legacy). * copilot core.ts / userDraftAdapter.ts: drop meta argument from saveAppDraft, loadAppDraftValue, write*Draft. Test assertions on getMeta dropped. Net: -22 typecheck errors, fewer moving parts, conflict modal works. EOF ) * refactor(drafts): remove dead endpoints + UserDraftDbSyncer.getLastSync The list_drafts and get_draft (own) routes were added during PR iteration and never wired up to any frontend caller — the editor overlay path uses the per-kind get-by-path getDraft query parameter, and the home page lists drafts via the per-kind list endpoints, not via /drafts. Drop both routes (+ sqlx caches + OpenAPI entries). UserDraftDbSyncer.getLastSync was a peep-hole for callers that never materialised — the per-tab lastSync map is only ever read by postSave internally, where the bookkeeping already lives inline. * refactor(drafts): extract DraftEditorModals trailer block The four editor routes (scripts/flows/apps/apps_raw) mounted an identical pair of trailer modals — DraftSyncConflictModal + OtherUsersDraftsModal — wrapped in the same guard chain and {#key path} remount. Lift the markup into one component; routes thread their itemKind, path, editPathFor, and loader callback. Pure markup extraction, no state ownership change. Drops the unused userStore import where the trailer was the only consumer. * refactor(drafts): UserDraft.useReactive — kill array-of-one boilerplate The script + flow routes both wanted a handle that re-keys when the URL path changes. UserDraft.use() can't do that (its opts getter is untracked), so each route hand-rolled the same useMany-array-of-one + proxy idiom: const handles = useMany(() => [{ kind, path: reactive }]) const handle = { get draft() { return handles[0]?.draft }, ... } Add UserDraft.useReactive(getSpec) that internally wraps useMany with a single spec and returns the stable proxy. Callers collapse to one line. * refactor(drafts): unify bootstrap suspension via armRestartOnFirstInteraction The flow and raw-app routes each rolled their own end-of-bootstrap resume: a 700ms setTimeout for flows and a templatePicker watcher with double-tick gating for raw-apps. Both are timing-fragile (the comments admit it) and drift from each other. armRestartOnFirstInteraction already existed in userDraftToast.ts for reset-to-deployed: keydown/input/pointerdown listeners (capture phase) that fire restartSync on the first real user touch, with a 5s belt-and-braces fallback. Export it and use it everywhere we'd previously have picked a magic number. For raw-apps this is a tiny behavioural change: the user's template choice now POSTs immediately (the pointerdown that picks the template also resumes sync, so the picker's onStart write rides the wake-up). Previously the choice only persisted on the user's NEXT edit. That's strictly better — navigating away preserves the choice now. * refactor(drafts): type App.draft_path; drop the as-any cast The audit asked for the three editors to converge on one draft_path injection pattern. For App and Flow, the in-builder $effect-mutates- the-store idiom is wedged into a shape that doesn't natively own the field — App's editor type genuinely has no draft_path so the writer had to cast through `as any`, and consumers downstream did the same. The minimum viable fix: declare draft_path on the local App type (it's already a field on the autosaved JSON). Lifting the writes upward into a route-side merger would mean restructuring the AppEditor mirror $effect and the FlowBuilder pathStore plumbing — larger change for the same shape, deferred to a follow-up. Flow already has the typed cast localised at one site. Will get the OpenAPI-level draft_path field as part of task 47 (drop as-any casts on backend overlay reads). * refactor(drafts): extract makeDraftAddLoad helper Four identical /add/+page.ts files differing only by the edit-route prefix. Lift the redirect into a factory, slim each entry point to two lines. * refactor(drafts): type UserDraftOverlay.other_drafts_users in the OpenAPI The backend response carried other_drafts_users on every get-by-path that supports the draft overlay, but the OpenAPI schema didn't declare the field. Each route had to cast the typed response to `any` to read it (and the sibling draft_saved_at), which obscured the real shape from the type system and rotted the discoverability of the draft surface. Add it to UserDraftOverlay. Frontend casts collapse to plain property reads in the three editor routes. * feat(drafts): list & open draft-only items for variables, resources, schedules, triggers For scripts/flows/apps the list and get-by-path endpoints already surface per-user drafts that have no deployed counterpart — that's what gates the home page from 404'ing on an AI-agent-created draft. Extend the same support to the other UserDraftItemKinds: Backend (list endpoints): - Add include_draft_only to ListVariableQuery, ListResourceQuery, ListScheduleQuery, StandardTriggerQuery (the latter covers the 11 trigger kinds via the generic TriggerCrud). - Append per-user draft rows whose path has no deployed row. Same gate as scripts/flows/apps: non-operators, page 0, no narrowing filters. Synthesis is per-kind: ListableVariable/Resource get field-for-field synthesis; ScheduleLight reads NewSchedule shape; Trigger<T> uses a best-effort JSON merge + serde_json::from_value (rows skipped on deserialize failure rather than failing the list). - Add draft_only: Option<bool> with sqlx(default) to each row type so it serializes as the column is opt-in. Backend (get-by-path endpoints): - get_variable, get_resource, get_schedule, get_trigger<T> fall back to fetch_draft_only when the deployed row is missing and the caller passed get_draft=true. Mirrors scripts/flows/apps. OpenAPI: - Shared IncludeDraftOnly parameter under components/parameters, wired into the 11 trigger list endpoints + listRawApps. Inline declarations on listVariable / listResource / listSchedules / listAzureTriggers. - draft_only field on ListableVariable, ListableResource, Schedule, TriggerExtraProperty. Frontend: - variables, resources, schedules, and the 10 trigger list pages (routes + 9 *_triggers) pass includeDraftOnly: true on the initial fetch and render <DraftBadge draft_only> on synthesized rows. Trigger pages got a sed/perl bulk update — pattern is the same across kinds. * fix(drafts): swap crypto.randomUUID() for the project's randomUUID helper crypto.randomUUID() is gated on a secure origin (HTTPS or localhost). Self-hosted Windmill instances often run on a bare HTTP origin or a LAN IP where the WebCrypto API is unavailable, so the /add redirect would throw before issuing the 307. Use the existing RFC4122 v4 helper in FlowChatManager that the rest of the codebase already imports for this exact reason. * fix(editor): leading-edge fire + max-wait cap on Monaco debounce The Editor debounced `onDidChangeModelContent` purely on the trailing edge — every keystroke rescheduled a 500ms timer, and uninterrupted typing held the bindable `code` prop stale until a pause. Stacked behind our 1.5s autosave debouncer that meant our clock didn't even start ticking until 500ms after the user paused, and the `code` binding never updated mid-burst for downstream consumers (lint, live preview, change listeners). Switch to leading + trailing + max-wait: * First keystroke of a burst fires `updateCode` synchronously, then stamps a wall-clock chain start. * Each subsequent keystroke (re)arms a trailing timer at `min(now + changeTimeout, chainStart + maxChangeTimeout)` — the cap is what makes continuous typing materialize at least once per maxChangeTimeout window instead of indefinitely. * When the trailing fires it resets the chain so the next keystroke after a pause is a fresh leading fire. New prop `maxChangeTimeout` (default 1000ms) sits next to the existing `changeTimeout` (default 500ms). Dispose path clears the chain stamp alongside the timer. * feat(drafts): wire Ctrl/Cmd+S to flush the pending autosave immediately Each builder already had a Ctrl/Cmd+S keybinding routed through a saveDraft() no-op left over from the LS-era — the comment said "persistence happens via the page-level UserDraft autosave" but the shortcut was the user's only way to actually force a save without waiting for the 1.5s debounce. Restore the intent. * UserDraftDbSyncer.flush({ workspace, itemKind, path }) — new method that re-submits whatever's queued in pendingSaveOpts with immediate: true. No-op when nothing's pending. * Editor.svelte.flushPendingChanges() — exposes a synchronous updateCode() with chain reset, so callers can drain Monaco's own trailing debounce before asking the syncer to flush. Without this step a Ctrl+S within ~500ms of typing would POST the pre-burst content. * ScriptBuilder.saveDraft() — editor?.flushPendingChanges() → await tick() → UserDraftDbSyncer.flush(). Toast on result. * FlowBuilder.saveDraft() — no direct Monaco ref (flows have many per-module editors); just flushes the syncer. Editor.svelte's new 1s max-wait cap means at most the last <1s of typing in a module Monaco won't be in this POST; it follows in the next autosave round. * RawAppEditor.handleKeydown — adds a 's' case that flushes before the focus guard, so the shortcut fires regardless of where focus is in the editor pane. * fix(drafts): low-code apps — drop spurious autosave on /edit + remount on Load from server Two bugs in low-code app editor (raw apps use a separate code path): 1. Every /edit visit looked like an autosave because loadApp() called UserDraft.discard('app', path, undefined). The comment claimed "this load doesn't POST" but discard always POSTs value: null server-side — that surfaced as a DELETE-my-draft on every page load AND a flash in the AutosaveIndicator. The discard was originally intended to wipe the in-memory cell so AppEditor remounts "fresh". But the path-change $effect upstream already sets app = undefined before each loadApp, which unmounts AppEditor and releases the handle's entry — so a remount via app = backendApp naturally starts with an empty handle. Drop the discard. 2. The conflict modal's "Load from server" called loadApp() but didn't remount AppEditor. Since AppEditor's stateApp is captured once at mount and doesn't react to prop changes, the editor kept showing the conflicting local edits even after a successful reload. Wrap the onLoadFromServer to await loadApp() then bump redraw to force a fresh mount. * feat(drafts): home-page Draft badge — show user-initial circles, drop the '+' The home-page Draft badge previously showed '+Draft' as a flat label. Add per-user awareness: up to 3 user-initial circles render to the left of the label, ordered alphabetically; with 4+ users we collapse to the first 2 + a '+N' overflow circle so rows stay compact. Backend: * New `DraftUserRef { username: Option<String> }` in windmill-types::user_drafts, re-exported from windmill-common so the list endpoints in scripts/flows/apps crates share one import path (windmill-types/windmill-common can't be reordered without a cycle). * ListableScript / ListableFlow / ListableApp gain a `draft_users: Option<sqlx::types::Json<Vec<DraftUserRef>>>` field. The list SQL adds a per-row subquery `SELECT json_agg(...) FROM draft d LEFT JOIN usr u ...` that aggregates the workspace users with a per-user draft at this path. NULL (no drafts) decodes to None; LEFT JOIN against `usr` lets orphaned drafts (user removed from workspace) still surface with username = None. * Synthesized draft-only rows set draft_users to a single-element vector with the authed user (those rows come from `email = $2`). OpenAPI: `draft_users` added to listScripts / listFlows / ListableApp response shapes as an array of `{ username }` with nullable username. Frontend DraftBadge: * Accepts `draft_users: { username?: string | null }[]`. Renders up to MAX_CIRCLES (3) initial circles; at 4+ users renders first 2 + a gray '+N' overflow circle. * Initials: 'john.doe'/'john_doe' → 'JD', 'alice' → 'AL', the legacy NULL-email row → '?'. * Color picked deterministically from a 6-entry palette so the same user gets the same circle color across rows. * Label is now just 'Draft' (dropped the '+'). 'Draft only' is unchanged. * Tooltip lists every user in full. ScriptRow / FlowRow / AppRow thread `draft_users` through their prop types and pass it to DraftBadge. * fix(drafts): suppress 'You have unsaved changes' banner when deployed baseline is null A brand-new variable/resource/trigger (no deployed row yet) has `getDeployed() == null`, but the caller's `show` prop is computed off `current != deployed` which is trivially true while the user types. Result: the banner appeared with 'Show diff' (no-op — the drawer early-returns on null deployed) and a 'Discard' that's semantically backwards (there's nothing to revert to). Gate `show` internally on `getDeployed() != null`. The check sits in the banner rather than each caller because every caller would otherwise need the same boilerplate guard. * fix(drafts): hide LocalDraftBanner when deployed and current match the DiffDrawer's compare Earlier I gated the banner on `getDeployed() != null`, but the user still saw it fire on entries where 'Show diff' opens to 'No changes detected'. That means `show` (the caller's coarse dirty check) flagged a difference the DiffDrawer treats as a no-op — typically toggle defaults (`false ↔ undefined`), removed empty arrays, or key-ordering noise that `cleanValueProperties + orderedYamlStringify` collapses. Replicate the drawer's comparison inside the banner: stringify both sides through the same pipeline and only render when the keys differ. A single `diffKey()` helper keeps the logic local; the catch-and-empty fallback survives a non-serializable side rather than throwing. * ui(drafts): nest user-initial circles inside the Draft badge Previously the circles sat alongside the Badge in a parent flex container; the result read as two separate UI elements. The Badge component already exposes its children as a snippet rendered inside its own flex row, so moving the circles into it makes them feel like part of the same chip. Knock-on tweaks: shrunk the circles from h-4/w-4 to h-3.5/w-3.5 so the badge stays compact, and tinted each circle's ring with the badge's indigo palette (instead of plain white) so the overlap reads as a deliberate stack rather than dots floating on top of the chip. * feat(drafts): drop the authed user's circle, mark own drafts with a '*' suffix Three tweaks to the home-page Draft badge: 1. Filter the authed user out of `draft_users` before rendering circles. The row already signals 'this user has a draft' via the asterisk (below), so a circle for them would be redundant noise. New `currentUsername` prop on DraftBadge — pass `$userStore?.username` from each row. The tooltip still lists every user (with `(you)` next to the authed one) so the full picture is one hover away. 2. The badge already showed whenever `is_draft || draft_users.length > 0` (per-user OR any-user). Spelled the rationale out in a comment — no logic change. 3. Append '*' to the displayed summary when `is_draft` is true. Falls back to `draft_path`/`path` when summary is empty so the marker never decorates an empty string. Threaded the same expression into ScriptRow / FlowRow / AppRow. Slice/overflow math now keys on the post-filter `otherUsers` list, so dropping the authed user doesn't silently shrink the visible count (e.g. 3 users incl. self → 2 circles, not 1 circle + a '+1' bubble). * feat(drafts): clone per-user drafts when forking a workspace `clone_workspace_data` clones every other workspace-scoped table on fork creation (resources, variables, scripts, flows, apps, raw apps, triggers, schedules) but quietly dropped the `draft` table. With per-user drafts that meant any open editor in the parent lost its pending edits the moment a fork was created — surprising and inconsistent with how forks treat the deployed surface. New `clone_drafts` mirrors the existing clone helpers: a single INSERT...SELECT into the target workspace, preserving `path`, `typ`, `value`, `created_at`, and `email`. The `email` FK targets `password.email` which is instance-scoped so it carries across workspaces without remap. `created_at` is preserved on purpose so the per-tab `last_sync` baseline lines up with the parent's timeline — otherwise the fork's next autosave would race a stale `last_sync` and trip the conflict modal on every cloned draft. Plain INSERT (not UPSERT) is safe because the fork target is empty at create time; no conflict against the partial unique indexes (`draft_pkey_with_user` / `draft_pkey_legacy`). The synthetic BIGSERIAL `id` PK is regenerated by the default so it stays out of the column list. * ui(drafts): pin the authed user to the first circle instead of hiding them Previously the authed user was filtered out of the circle row entirely on the theory that the row's '*' suffix already signalled 'this user has a draft'. New requirement: they should always lead the circle row when they have a draft so the visual half of the signal lines up across rows (consistent leading-slot identity, easy scan). Switch from a filter to a sort: `orderedUsers` finds the authed user in `draft_users` and splices them to index 0; everyone else keeps the backend's alphabetical order behind. Slice/overflow math now keys on `orderedUsers`, which guarantees the authed user never falls into the '+N' bubble — they're at position 0 and the slice keeps the head. The popover's '(you)' annotation moves to the circle's title attr too, so hovering the leading circle confirms the identity. * feat(drafts): drop draft_only column from script/flow/app Drafts now live in the `draft` table exclusively — `draft_only` stubs in script/flow/app are redundant. Migration `INSERT INTO draft ... ON CONFLICT (workspace_id, path, typ) WHERE email IS NULL DO NOTHING` so real per-user drafts already at the same path are preserved; only rare stubs that lost their draft get a synthesised workspace-level row. Stubs are then deleted (FKs cascade to *_version) and the column is dropped. List endpoints keep a synthesised `draft_only: true` on rows sourced from the draft table itself (sqlx default on the struct field). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * ui(drafts): surface draft state in AutosaveIndicator instead of toast+auto-modal The "Loaded your saved draft" toast and the auto-opening OtherUsersDraftsModal both surprised users on every editor mount. Move both signals into the AutosaveIndicator label: "Loaded from draft" or "Others are working on this {kind}" (priority) sits where Saving/Saved do, with a one-shot light-green flash behind the indicator that fades to transparent. Saving/Saved still win when they fire. The popover gains a "See others' drafts" button that flips the modal open on demand; the modal itself is now externally controlled via a bindable \`isOpen\` threaded through DraftEditorModals. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * ui(drafts): per-user View JSON / Fork actions in DraftBadge popover Hover popover used to be a plain text list of usernames. Now each row gets a colored circle icon + name + "(you)" for the authed user, and every OTHER user's row carries View JSON / Fork buttons mirroring the OtherUsersDraftsModal. For draft-only entries owned solely by the authed user, the popover ends with "Only you can see this {kind}" so the row's privacy is obvious. ScriptRow / FlowRow / AppRow thread workspace + itemKind + path + editPathFor through; AppRow switches between app / raw_app on app.raw_app. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * nit * fix(drafts): clone only the forker's per-user drafts on workspace fork clone_drafts copied every user's drafts, but only the forker gets added to the fork's usr table. Drafts owned by absent users LEFT-JOIN to NULL in the home page's draft_users aggregate, surfacing as multiple legacy-style rows at one path and crashing the popover with each_key_duplicate. Filter the clone to email = forker OR email IS NULL, and key the popover's #each by index defensively so future legacy collisions can't crash the page either. Also re-adds `draft_only: None` to NewScript/CreateFlowBody literals in tests — the auto-generated windmill-api-client still carries the field and the previous commit dropped them too aggressively. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(drafts): always populate other_drafts_users in maybe_overlay_draft Reset-to-deployed reloads the deployed payload with get_draft=false, which made the backend return other_drafts_users=[]. The route then reassigned otherDraftsUsers to the empty list, dropping the count to 0 and hiding "See others' drafts" in the AutosaveIndicator popover — but the other users' drafts hadn't actually gone anywhere. Fetch the list independently of get_draft so the popover stays accurate across reset reloads. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(drafts): alert user when their draft is older than the latest deploy Open a modal on editor mount when the per-user draft was saved before the latest deploy at the same path — i.e. a teammate deployed a new version while this user's draft was sitting. Two choices: discard the stale draft and pick up the deploy, or keep editing the older draft. DraftEditorModals computes the staleness from the timestamps each route threads in (script.created_at, flow.edited_at, app_version.created_at) and the "Load latest deploy" callback reuses the route's existing reset-to-deployed logic. Wired for script / flow / app / raw_app editors; trigger / resource / variable drawer editors follow a different pattern and aren't covered here. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(drafts): deploy only wipes the deployer's draft, not everyone else's Script / flow / app deploys ran an unconditional DELETE on every draft at the path, so a teammate's deploy silently destroyed any other user's pending draft. After the wipe, the other user's tab kept auto-saving — re-creating the row at a NOW timestamp newer than the deploy — and StaleDraftModal never fired because draft_saved_at had been bumped past the deploy. Filter the DELETE to email = deployer (plus the legacy NULL row), so other users' drafts persist and the stale-draft prompt actually fires on their next reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(drafts): surface save failures in AutosaveIndicator instead of pretending Saved postSave caught network errors with `console.error` and let the runner finish normally. The indicator read the saving → none transition as a successful save and flashed "Saved" even when the request had thrown. Track failed keys in a SvelteMap, expose `'failed'` as a new UserDraftSyncState, render "Save failed" in red with a CloudOff icon. Failure clears on the next successful save for the same key, or when recordRemoteSync seeds a fresh authoritative timestamp. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(drafts): surface 'Save failed' inside the AutosaveIndicator popover too The popover used to repeat the cheerful "All changes are saved as a draft on the server..." copy even when the inline label said "Save failed", which read as contradictory. Add a red, text-xs warning at the top of the popover body when the sync state is `failed`, explaining that the latest edits didn't reach the server and that editing again retries the save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(drafts): surface the actual error message in the AutosaveIndicator popover Replace the generic "your latest changes did not reach the server" copy with the real failure detail. The syncer now stores the extracted message in the failures map (formatSaveError walks body / message / statusText) and exposes it via the state handle's `failureMessage` getter. Popover renders it in red, monospaced, scrollable so a long server traceback doesn't blow out the popover. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(drafts): suppress Saving/Saved indicator during a reset-to-deployed discard A `value: null` POST is a discard, not a save, but it ran through the same runner the indicator watched — so resetting to deployed flashed "Saving..." → "Saved", reading as "your draft just landed" while we were actually wiping it. Track in-flight discards in a SvelteSet, expose a distinct `'discarding'` UserDraftSyncState, and the indicator stays quiet for it: no spinner, no label change, and the `discarding → none` transition deliberately skips the "Saved" flash. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * Revert "fix(drafts): suppress Saving/Saved indicator during a reset-to-deployed discard" This reverts commit |
||
|
Ruben FiszelandClaude Opus 4.8
|
066d7a4726 |
block operators from inline preview job execution (#9572)
`POST /api/w/{workspace}/jobs/run_inline/preview` ran request-supplied
code inline (in-process, e.g. DuckDB) but was missing the operator
authorization guard that its sibling `/jobs/run/preview` enforces. An
authenticated operator — the most restricted role, which must not run
preview jobs — could execute arbitrary code in a single request
(file read/write, and OS command execution via the DuckDB `shellfs`
extension when worker egress is available).
This is the incomplete-fix residual of CVE-2026-22683 / GHSA-9q9g-rp9x-244h,
whose v1.615.0 patch covered the entity-CRUD endpoints but left this
direct inline-exec sink uncovered.
Add the same `is_operator` guard from `run_preview_script`. Audited the
rest of the preview/inline arbitrary-code endpoints (run_preview_script,
run_bundle_preview_script, run_preview_flow_job, the wait_result
wrappers, run_dynamic_select inline variant, dependency jobs) — all
already carry the guard. The `run_inline_script_by_path`/`by_hash`
endpoints run deployed scripts (operator-allowed, scope-checked) and
correctly remain ungated.
Fixes WIN-2043 (GHSA-pp5h-96x3-3wqq).
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Ruben FiszelandClaude Opus 4.8
|
317a8629d1 |
fix(ai): enforce resource authz when loading MCP tools in agent worker (#9571)
* fix(ai): enforce resource authz when loading MCP tools in agent worker Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test: describe MCP authz test invariants instead of drafting history Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Ruben Fiszel
|
d0aaba0f16 |
require pinned sha for inline raw_code in viewer app run mode (#9570)
* fix: require pinned sha for inline raw_code in viewer app run mode * fix: gate rd_string import behind parquet feature to fix oss build * fix: also require pin for raw_code with app_script id in viewer run mode |
||
|
Ruben FiszelandClaude Fable 5
|
2aab35245c |
fix(flows): skip_if evaluates wrong previous_result during retry (#9547)
* fix(flows): evaluate skip_if against previous step result on retry Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * docs: narrow retry previous_result restoration comment Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Fable 5
|
cf9ad54181 |
feat: workspace protection rule to restrict anonymous app deployment (#9509)
* feat: restrict anonymous app execution mode to admins Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * test: add regression test for anonymous app admin gate Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat: gate anonymous app mode behind workspace protection rule Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix: lock app row on anonymous-mode check, fail closed while rules load Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Fable 5
|
6fb6249cf9 |
refactor: ungate cascade producers, squash pipeline migrations
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> |
||
|
Ruben Fiszel
|
0009050127 | merge | ||
|
fad1a549d9 |
feat(otel): connect jobs to the inbound distributed trace (#9456)
* feat(otel): propagate inbound W3C traceparent to job spans Capture the inbound traceparent header at the run endpoints (WebhookArgs::to_args_from_format) into a reserved _wm_traceparent arg key (gated on OTEL_TRACING_ENABLED), riding the args jsonb like _ENTRYPOINT_OVERRIDE. At pickup, create_span_with_name attaches a span link from the job's worker span to the originating distributed trace, so a job triggered by an instrumented service is connected to the caller's trace while keeping its UUID-derived trace id (trace-by-job-id unaffected). The link/parse logic lives in the EE otel modules; this OSS side only captures the header and calls the (no-op outside EE) hook. Companion EE PR required. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore: bump ee-repo-ref to inbound-trace-propagation EE branch Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * docs(agents): don't attribute work to specific customers in repo content Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(otel): relocate job + script spans into the inbound trace Builds on the captured _wm_traceparent: the worker job span is re-parented on the inbound caller context, the script subprocess's TRACEPARENT env is the inbound context (so its spans join the caller's trace), and the context is propagated to flow steps so the whole flow relocates. Carried to the worker via a new LogContext.inbound_traceparent field. Non-inbound jobs are unchanged. Adds a relocation integration test. Companion EE PR required. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore: bump ee-repo-ref to inbound-trace-propagation relocate commit Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(otel): harden inbound traceparent capture Address review feedback: - strip any caller-supplied _wm_traceparent from args/extra before stashing the header-captured value, so the reserved key is Windmill-controlled only - valid_w3c_traceparent: reject version ff and require lowercase hex, so we don't forward an inbound header that downstream OTel parsers would reject - clarify that the capture helper does not validate the W3C format (done at use) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore: update ee-repo-ref to 2c7964460327fab5e3a27c0f74b8d6f26ab7f79a This commit updates the EE repository reference after PR #604 was merged in windmill-ee-private. Previous ee-repo-ref: 8fc04fb105dc49769205f7174d551a0d134d1bec New ee-repo-ref: 2c7964460327fab5e3a27c0f74b8d6f26ab7f79a Automated by sync-ee-ref workflow. --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com> |
||
|
hugocasaandClaude Opus 4.8
|
f2f0812a04 |
feat(flows): opt-in to include the stopping step's result in early-stop errors (#9446)
* feat(flows): early stop can include the stopping step's result in the raised error
When a step uses Early Stop with "Raise an error message if stopped", the
flow result was entirely replaced with a static error object
({"error": {"name": "EarlyStopError", "message": "..."}}), discarding the
stopping step's own output. This made it impossible to stop+fail a flow
while preserving the data the step produced (e.g. an API that returns
HTTP 200 with a userErrors payload).
Add an opt-in `error_include_result` flag on StopAfterIf. When enabled on
the raise-error path, the raised payload becomes
{"error": {...}, "result": <step result>} instead of dropping the result.
Default is false, so existing behavior is unchanged. The option is threaded
through the worker's stop-after-if handling (including stop_after_all_iters_if
for loops/branchall) and exposed in the flow editor's Early Stop panel.
Fixes WIN-2012
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* test(flows): cover early-stop error_include_result payload shaping
Add a regression test asserting that a step using Early Stop with a raised
error message and error_include_result=true fails the flow while preserving
the step output as {"error": {..}, "result": <step result>}, and that with
the flag off the result is the bare {"error": {..}} object.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* refactor(flows): nest early-stop step result inside the error object
Embed the stopping step's result under `error.result` rather than as a
top-level sibling of `error`. This keeps the flow result shape as
`{ "error": { .. } }` — identical to a normal error — so consumers that
key off the top-level shape (single `error` key) keep working, while the
data is still preserved for those that look inside the error object.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(flows): always include the stopping step's result in early-stop errors
Drop the opt-in `error_include_result` gate. Since the step result is nested
inside the error object (`error.result`), the top-level result shape stays
`{ "error": .. }` — identical to a normal error — so consumers that detect or
parse failures by the top-level shape are unaffected. Gating it added schema
surface, plumbing, and a UI toggle for no real compatibility benefit.
Now, whenever a step early-stops with a raised error message, the flow fails
and the raised error embeds the stopping step's own result under
`error.result` (aggregated iteration results for loops/branchall). This
reverts the `StopAfterIf.error_include_result` field, its threading, the
OpenAPI/generated-client surface, and the editor toggle; the "Raise an error
message" tooltip now notes that the step result is included.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(flows): gate early-stop result inclusion behind opt-in flag
Re-introduce the per-step `error_include_result` flag (default off) instead
of always embedding the step result. Although nesting the result under
`error.result` keeps the result *shape* backward-compatible, it does not
address data exposure: a failed flow's result is propagated to synchronous
webhook callers, the flow's failure module, and the workspace/global error
handler (commonly a Slack/email/outbound-webhook notifier). Always including
the step output would surface previously-redacted intermediate data to all of
those sinks for every existing error-stop flow.
Gating keeps the existing behavior (bare `{ "error": .. }`) as the default and
only embeds `error.result` when the flow author explicitly opts in, matching
the original issue's intent.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(flows): omit error_include_result when false; refresh generated prompts
- Add `skip_serializing_if = "is_false"` to `StopAfterIf.error_include_result`
so serialized flows are byte-identical when the flag is off. Fixes the
`flowmodule_serde` round-trip test (cargo_test) and avoids churn on existing
flows.
- Regenerate `system_prompts/auto-generated/` and `cli/src/guidance/skills.gen.ts`
for the new OpenFlow `error_include_result` property. Fixes check-freshness.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* test(flows): cover error_include_result for the loop "stop after all iters" path
Add a regression test for the stop_after_all_iters_if branch, where `nresult`
already holds the aggregated iteration results — confirming `error.result`
carries each iteration's output (distinct from the per-step fallback path).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
hugocasaandClaude Opus 4.8
|
468aa230e5 |
refactor: resolve workspace imports via /f/,/u/ not $f/,$u/ aliases (#9438)
* refactor: resolve workspace imports via /f/,/u/ not $f/,$u/ aliases Keep the CLI managed tsconfig.wmill.json / `refresh tsconfig` / Deno import-map QoL from #9378, but re-key it on the existing /f/,/u/ workspace paths instead of the new $f/,$u/ specifiers. Verified /f/,/u/ resolves in tsc, Bun, Deno, the in-app ATA editor, and the worker, so the $-prefixed alias added no value. Drop the $f/,$u/ handling from the parser, dep-map, deno_executor, bun loaders, ATA, relative_imports and monaco paths; revert the windmill-parser-wasm-ts bump (1.714.0 -> 1.695.0). Also fold in the cli/package-lock.json sync for the already-committed pg-gateway dependency. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix: drop duplicate relative-path check and restore rustfmt formatting Follow-up cleanups to the previous commit's full-file reverts, which restored pre-#9378 state that main had since improved: - relative_imports.ts: remove the redundant duplicate d.startsWith('/') (pre-#9378 had it; #9378 had repurposed that line, so main has no dup). - windmill-parser-ts/src/lib.rs: restore the multi-line new_source_file(...) formatting required by backend/rustfmt.toml (the single-line revert would fail `cargo fmt --check`). Now differs from main only by the $f//$u/ removal. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
hugocasaandClaude Opus 4.8
|
220cd35cf7 |
feat: support $f/ and $u/ import path aliases for scripts (#9378)
* feat: support $f/ and $u/ import path aliases for scripts $f/ and $u/ are local-friendly aliases for the absolute workspace import paths /f/ and /u/. Unlike the /-prefixed form (which local tools treat as a filesystem-root path), the $-prefixed form is a bare specifier that can be remapped via tsconfig paths / Deno import maps, so the same import resolves on the Windmill worker and in a local editor. - worker: recognize $f//$u/ in the Deno import map and both Bun loaders - dep-map/parser: normalize $f/->f/, $u/->u/ for lockgen + dep tracking - cli: emit $f/$u path aliases in generated tsconfig.json / deno.json - frontend: ATA + Monaco paths resolve $f//$u/ type hints in the editor Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(cli): split generated tsconfig into managed + user file with refresh command Mirror the AGENTS.cli.md/AGENTS.md prompts model for the IDE tsconfig so the recommended settings can evolve without ever clobbering user customizations: - tsconfig.wmill.json: wmill-managed, always refreshed, holds recommended compilerOptions incl. the $f/$u path aliases (Deno: import_map.wmill.json) - tsconfig.json: user-owned, created once, just extends the managed file; warn (never auto-edit) when an existing one doesn't reference it - add 'wmill refresh tsconfig'; init generates it unconditionally (no longer gated behind resource-type namespace / a bound workspace) - regenerate CLI guidance docs for the new subcommand Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(cli): address PR review on $f/ tsconfig generation - handle existing deno.jsonc so we don't shadow it with a new deno.json (P1 identified by cubic) - fix the bun-types hint that pointed users at the managed do-not-edit tsconfig.wmill.json; tell them to install + re-run 'wmill refresh tsconfig' - document the .ts-extension-only local-resolution limitation (cross-flavor .bun.ts/.deno.ts/.fetch.ts scripts won't resolve in a local editor) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(cli): warn when a project's tsconfig isn't wired to tsconfig.wmill.json Mirror the prompts freshness check for the managed tsconfig so users with an existing setup actually discover they're missing $f//$u/ resolution: - embed a version hash in tsconfig.wmill.json (excludes the env-dependent bun-types 'types' entry so it doesn't false-positive) - add warnIfTsconfigStale to the main.ts freshness hook, gated identically to the prompts check (skips init/refresh/help/version). When a tsconfig.json exists it warns one line (stderr) if the managed file is missing, not referenced via extends, or out of date; silent for non-TS projects. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * refactor(cli): make tsconfig setup equivalent to prompts (auto-wire + stale-only) Unify the two managed-file systems so they behave identically: - auto-wire an existing unlinked tsconfig.json/deno.json on init/refresh (add extends / importMap; merge into an array extends), instead of only warning. Parses JSON and falls back to a warning when it can't round-trip (JSONC comments, or a conflicting deno imports/importMap) — never corrupts. - narrow warnIfTsconfigStale to stale-only, gated on the managed file existing, exactly like warnIfPromptsStale: it no longer nags about a missing or unlinked tsconfig.json, so a deliberately-custom/unlinked setup stays silent and a not-yet-initialized project isn't bothered. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(cli): place tsconfig.wmill.json first in extends to preserve user base config Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(cli): migrate legacy tsconfig and require consent for custom configs Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * refactor(cli): align prompts wiring to the same consent model Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore(cli): bump windmill-parser-wasm-ts to 1.714.0 for $f/ $u/ aliases Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(worker): resolve $f/ and $u/ in deno lock generation Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test: narrow relative-imports lock-gen guard to deno import-map failure Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore(cli): sync bun.lock with windmill-parser-wasm-ts 1.714.0 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(cli): warn when a custom tsconfig's paths would shadow $f/ $u/ aliases Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Opus 4.8
|
8053266f88 |
fix(mcp): resolve MCP resource token via caller RLS + SSRF-guard url (#9428)
* fix(mcp): resolve MCP resource token via caller RLS + SSRF-guard url Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(mcp): clone user_db for oauth2 refresh and drop advisory ids from comments Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(mcp): disable redirects on MCP client to prevent SSRF bypass Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Opus 4.8
|
89a7a37776 |
fix(backend): authorize single-job read endpoints by job/flow visibility (#9416)
* fix(backend): authorize single-job read endpoints by job/flow visibility Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(jobs): share read links + cached access checks for run visibility - Cache the job read-access RLS probe (size-bounded LRU keyed by the caller's authz-relevant identity + job id; no TTL since job-side inputs are immutable). - Inherit visibility along the full parent_job chain so any flow you can see lets you read its (deeply nested) steps. - Share read links: GET /jobs/job_view_token/{id} mints a stateless HMAC(workspace_key, job_id) token (only if the caller can read the job); the token grants an authenticated member read of that job and its flow subtree via a ?view_token query param or X-View-Token header. Run page gains a Share button and honors a ?view_token link. - Denied-but-existing reads now return 403 with guidance to request a share link (vs 404 for non-existent), and the run page renders that case with instructions. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(jobs): address PR review — scope-tag check on mint, constant-time view-token verify - P1 (Codex): get_job_view_token now enforces the caller's if_jobs:filter_tags scope before minting, so a tag-scoped token can't mint a transferable link for a job outside its tags. Adds a scoped-token regression test (allowed + denied). - Constant-time view-token verification (HmacSha256::verify_slice) instead of comparing hex strings (Claude/Pi nit). - get_completed_job_result: an authed reader passing an invalid suspended-secret triple now falls through to the normal visibility gate instead of erroring out (Claude nit); unauthenticated callers still rejected. - Length-prefix the read-access cache key fields so no input values can collide. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * docs(api): add job_view_token to openapi spec; use generated client in run page Addresses Codex review nit: the new GET /jobs/job_view_token/{id} endpoint was missing from openapi.yaml (the source the frontend client is generated from). Adds the path + operationId getJobViewToken, and switches the run page's Share button from a raw fetch to JobService.getJobViewToken. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(frontend): carry view_token on share-link downloads Addresses Codex review: download actions bypass the request interceptor that adds X-View-Token (downloadViaClient uses raw fetch; cookie-mode downloads use plain hrefs), so a share-link viewer got 403 downloading logs/results/args. Append the view_token query param to the job download paths (result/logs/args/flow-all-logs) via a new appendViewToken() helper, covering both client-fetch and href modes. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(jobs): enforce tag scope in require_job_read_access (view-token use side) Addresses Codex P1: the view_token use-side bypassed if_jobs:filter_tags on handlers that don't tag-filter their data query (result_by_id, get_flow_job_debug_info, get_otel_traces) — a tag-scoped token could use someone else's valid share token to read out-of-scope job data. Move the tag-scope check into require_job_read_access (runs before any created_by/view_token/RLS grant), so it applies uniformly to every gated handler; removes the now-redundant explicit check in get_job_view_token. Adds a use-side regression test (scoped token + valid out-of-scope view_token denied on otel/result_by_id; in-scope still allowed). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(frontend): include workspace in share read link Addresses Codex P1: the copied share URL omitted the workspace. The token is signed with the run's workspace key and the logged layout only switches $workspaceStore when the URL carries workspace=, so a recipient whose persisted active workspace differs would open the link against the wrong workspace and the token would fail validation. Pin workspace= alongside view_token in the link. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(jobs): authorize get_result_maybe get_started branch for queued jobs Addresses Codex P1: get_completed_job_result_maybe only gated when a completed row existed; with ?get_started=true a non-reader reached the fallback branch and got started:true for a running private job. Now fetches created_by and authorizes (created_by/view_token/RLS, or anonymous for unauth) before disclosing running-state; a non-existent job still returns started:false (leaks nothing). Adds a regression test with a queued (no completed row) private job. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Opus 4.8
|
73edebc833 |
fix(backend): route //native TypeScript previews to native workers (WIN-2007) (#9407)
* fix(backend): route //native TypeScript previews to native workers Previewing a TypeScript script carrying the `//native` annotation was pushed with `language = bun` (what the editor sends), so the job was tagged `bun` and routed to a regular bun worker. A native-mode worker neither matches the `bun` tag nor accepts a non-native `script_lang` (worker.rs rejects with "cannot execute non-native job with language 'bun'"), so previewing a `//native` script on a native-only worker setup failed — even though the deployed version of the same script runs fine as `bunnative` / tag `nativets`. `push` now reconciles the preview language with the `//native` annotation for `JobPayload::Code`, mirroring the deploy-time logic in `worker_lockfiles`: `bun` + `//native` is promoted to `bunnative` (tag `nativets`), and `bunnative` without `//native` is demoted back to `bun`. This makes a preview run exactly like the deployed script would, and covers every preview entry point (run_preview_script, inline preview, codebase preview) since they all go through `JobPayload::Code`. Adds regression tests asserting the queued job's `script_lang`/`tag` for all four (declared language × annotation) combinations. Fixes WIN-2007 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore(backend): add sqlx cache for preview_native_tag test query The regression test's `sqlx::query!` for `v2_job` (tag, script_lang) needs a cached entry so `SQLX_OFFLINE=true` CI compiles it. Adds exactly one new cache file; no existing (OSS or EE) caches removed. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(backend): trim preview native-tag tests to the essentials Keep the core regression (bun + //native → bunnative/nativets) and the guard that plain bun previews are unaffected. Drop the two bunnative- declared cases, which only re-verified the mirrored demote logic and weren't the reported issue. The shared query is unchanged, so the sqlx cache stays valid. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Ruben Fiszel
|
fc5fe1e755 |
Merge branch 'main' into feat/asset-graph-view
# Conflicts: # backend/ee-repo-ref.txt # frontend/src/lib/components/ScriptEditor.svelte |
||
|
hugocasaandClaude Opus 4.8
|
b0c3b01d31 |
fix(cli): preserve user drafts on sync push and permissioned-as (#9381)
CLI deploys (sync push, set-permissioned-as) went through the same create/update endpoints as a UI "deploy from draft", which delete the draft at that path. That silently wiped teammates' in-progress drafts on every push. Add a transient skip_draft_deletion deploy flag (mirroring deployment_message) that the CLI sets; the backend then skips the DELETE FROM draft for scripts, flows, and apps. UI deploys are unchanged. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Ruben Fiszel
|
e5649c393c | merge | ||
|
Ruben Fiszel
|
42d2121af9 |
fix(queue): duration-weighted workspace fairness signal (#9329)
* fix(queue): bump EE ref to include worker_ping fairness signal The current ee-repo-ref.txt pointed to 31cda7c (an unrelated merge commit on the asset-graph-view-ee branch) instead of ddc9e80, which contains the workspace-fairness fix that switches the active-share signal from v2_job_queue.running=true to worker_ping. As a result cloud was still computing overload off the legacy signal, so a workspace with many in-flight/suspended flows (lancom01-prod, with 799 suspended flows × 3 v2_job_queue bookkeeping rows each = 2397 running-true rows) was flagged as 95% of cluster activity despite consuming zero worker slots. Bumping to ddc9e80 picks up the worker_ping-based signal, which naturally excludes (a) suspended jobs (no worker pinging them), (b) zombie running-rows from dead workers, and (c) flow/flownode orchestration rows that never run on a worker in the first place. * test(queue): seed v2_job rows + realistic durations for fairness helpers The new duration-weighted fairness algorithm joins v2_job_queue and v2_job_completed to v2_job for the `kind` filter (excluding flow bookkeeping) and reads `duration_ms` for the completed contribution. Update the test helpers to mirror that schema: * `insert_completed` now inserts a matching v2_job row (kind=script) and writes `duration_ms = 1000` with a 1-second [started_at, completed_at] interval, so each completed row contributes ~1 worker-second when fully inside the refresh window. * `insert_queued` likewise pre-inserts v2_job, sets `started_at` to NOW() - 1s when running=true (so running rows contribute ~1 worker-second by the time the refresh runs), and seeds v2_job_runtime.ping so the running side accrues real-time worker seconds (the algorithm bounds end-of-interval by ping). The zombie/suspended insert helpers are intentionally left without v2_job rows — the new algorithm's INNER JOIN excludes them, so they still correctly contribute zero worker-seconds. * chore(queue): bump EE ref to duration-weighted fairness algorithm Companion to windmill-ee-private#<TBD>: switch the EE workspace fairness aggregation from a count-based UNION (worker_ping snapshot + v2_job_completed count) to a worker-seconds aggregation sourced directly from v2_job_queue and v2_job_completed, with kind/suspend filters mirroring handle_zombie_jobs and per-row defenses against zombie inflation on both halves. * chore(queue): bump EE ref for fairness perf fix (inline window_start) * chore(queue): bump EE ref for fairness perf rewrite (driver-side flip) * update ee ref |
||
|
Ruben Fiszel
|
a6c51b146d |
ignore flaky fairness regression tests in CI (#9328)
`fairness_ignores_zombie_running_rows` and `fairness_ignores_concurrency_suspended_rows` panic intermittently in CI (both Linux and Windows runs). Mark them `#[ignore]` until the underlying flakiness is resolved. |
||
|
Ruben Fiszel
|
896add0350 | fix: improve workspace fairness | ||
|
Ruben Fiszel
|
e7031ae826 | Merge branch 'main' into feat/asset-graph-view | ||
|
Ruben Fiszel
|
8bf7fd2c92 | feat(queue): stochastic admission + EE availability of workspace fairness algorithm (#9321) | ||
|
hugocasa
|
268a6a7742 | fix: early return should consider failure_module result (#9241) | ||
|
Ruben Fiszel
|
55675ceb62 | feat: pipeline // tag and // retry annotations + dispatch_event log | ||
|
hugocasa
|
2db1c0a1fc | fix: early return should consider failure_module result (#9241) | ||
|
Ruben Fiszel
|
69dd3cd4a5 | chore: merge main into asset graph view | ||
|
Ruben Fiszel
|
631582f6e3 | Merge branch 'main' into feat/asset-graph-view | ||
|
Ruben FiszelandClaude Opus 4.7
|
4b1bea8aed |
fix: enforce auth guards on app component preview execution (#9235)
* fix: enforce auth guards on app component preview execution Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: guard previewed runnable path and worker tag in app preview Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: validate app_script id ownership and keep root push isolation Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * refactor: scope app preview guards to operator check + referenced runnables Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: require jobs:run scope and tag check on app preview (apps:run escalation) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Opus 4.7
|
bc67842f7a |
feat: reap abandoned AND-join slots after a TTL (default 60d, per-slot)
join_pending_inputs slots are normally cleared when the join fires; partial slots whose inputs never all arrive (upstream removed/renamed, one-off dynamic partition key, permanent skew) would otherwise leak. windmill_queue::asset_dispatch::reap_stale_join_slots, called from the monitor's delete_expired_items loop, deletes a (workspace, subscriber, partition) slot only when its MOST RECENT row is older than JOIN_SLOT_TTL_SECS (60d) — per-slot, never per-row, so a legitimately slow join is not corrupted mid-accumulation. Conservative default; per-join configurable TTL via the annotation is a planned follow-up. Test covers stale-reaped / fresh-kept / mixed-slot-kept. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Opus 4.7
|
3d0558f937 |
test: fuller partitioned join + multi-hop pipeline coverage
Exercises a complex pipeline combining options end to end: two partitioned producers fanning into a // trigger all join, then a multi-hop downstream chain. Asserts the resolved partition propagates unchanged at every hop, chain depth increments per hop, the AND barrier fires exactly once, and a second partition opens an independent slot with no cross-partition bleed across the whole graph. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Opus 4.7
|
d81f9737f6 |
test: AND-join fires once under concurrent upstream completion
Regression for the check-then-act race fixed by the advisory-locked transactional gate: releases N producer dispatches simultaneously via a barrier and asserts the AND subscriber is pushed exactly once and the slot is cleared. The invariant holds for the correct gate regardless of interleaving; a non-atomic regression fails it. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Opus 4.7
|
99e81c5b50 |
feat: apply opt-in debounce to asset-cascade subscriber dispatch
Stage E3. fetch_subscribers now also returns debounce_s; push_subscriber builds real DebouncingSettings (delay + a (subscriber, partition) key, so distinct partitions never collapse and latest-in-window falls out) instead of ::default() when the edge opted in. Default stays no-debounce (fan-out — the prior deliberate behaviour, now overridable rather than reversed). Wiring test asserts the dispatched job carries the configured window/key and an undebounced edge carries none. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Opus 4.7
|
124ccf5509 |
feat: AND-join barrier for partitioned pipeline subscribers
Stage D: a // trigger all subscriber no longer fires on any input. New
join_pending_inputs slot table keyed (workspace, subscriber, partition);
fetch_subscribers now returns join_all and the dispatch loop records each
partition-bearing input arrival, pushing the subscriber once only when
every partition-bearing input it declares is present for that partition.
Per-partition slots, cleared on fire (re-accumulate, no double-fire),
skew-immune (unlike debounce). Case-3 guard: an unpartitioned producer or
a reference (non-{partition}) input never fires a partitioned join.
Integration test covers wait/fire/isolation/no-double-fire.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
||
|
Ruben FiszelandClaude Opus 4.7
|
a888ef27f1 |
feat: resolve pipeline partition at job execution time
Stage C: in handle_code_execution_job, once the script content is loaded, parse the // partitioned annotation (free here) and resolve the concrete partition once — schedule fire-time (scheduled_for anchor, not wall-clock) for time kinds, triggering payload for dynamic. The value is injected into the in-memory args the body sees (via a shadowed job clone) and persisted back to v2_job.args so dispatch_asset_triggers propagates the same value down the cascade. Already-set (explicit/backfill/cascade) partitions are never re-resolved (run identity immutable); unresolvable partitioned runs fail with a clear error. Integration test exercises the full worker loop + cascade propagation. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> |
||
|
Ruben FiszelandClaude Opus 4.7
|
dfeed9c5c2 |
fix: actionable error when a custom_path is taken by an app in another workspace (#9190)
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
 
|
52960ca30a |
fix: reset parent_hash in auto_parent when all versions at path are archived (#9172)
* fix: reset parent_hash in auto_parent when all versions at path are archived * test: regression test for auto_parent with all versions archived --------- Co-authored-by: windmill-internal-app[bot] <1429786+windmill-internal-app[bot]@users.noreply.github.com> Co-authored-by: Ruben Fiszel <ruben@windmill.dev> |
||
|
Ruben Fiszel
|
b37e6830fe | merge | ||
|
Ruben Fiszel
|
f414ffc484 | fix: never mark failure/trigger/approval scripts as auto_kind=lib (#9168) | ||
|
Ruben Fiszel
|
638b947d00 | test: cover asset-trigger dispatch end-to-end through worker | ||
|
Ruben Fiszel
|
33bf01b627 |
fix(python): preserve strings containing Infinity/NaN in result JSON (#9149)
* fix(python): preserve strings containing Infinity/NaN in result JSON * test(python): add sanity checks for Infinity/NaN in results * refactor(python): use string-aware regex callback for single-pass cleanup * refactor(python): compact regex + handle backslash-escape parity * perf(python): short-circuit cleanup when no Infinity/NaN/NUL in result |
||
|
|
d666e8431c |
feat: read-only flag on API tokens (#9144)
* feat: read-only flag on API tokens, orthogonal to scopes Add a per-token `read_only` boolean set at creation time. When true, the token can only call HTTP methods classified as Read (GET/HEAD/OPTIONS). Mutating methods and job-run actions are rejected with 403, regardless of which scopes are attached. Surfaced as a prominent toggle in the standard token-creation flow and a discreet `2xs` toggle in MCP mode (where users often want write access, so we don't bias them toward enabling it). MCP enforcement: read-only tokens hide all script/flow/hub tools from `list_tools` and only see endpoint tools whose method is GET, and the runner rejects `call_tool` on anything mutating. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: review fixes for read-only token flag - Exempt /api/mcp/* and /mcp/* paths from the read-only middleware check. MCP transport runs over POST (streamable HTTP / SSE), so otherwise the middleware would 403 every MCP request before the runner could enforce read-only at the tool-call level. - Tighten is_endpoint_read_only to GET only, matching the read_only_hint that create_endpoint_annotations actually emits. - Add unit test for check_read_only_for_route covering GET/HEAD/OPTIONS, mutating methods, and run paths. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * chore: bump ee-repo-ref to read-only-trigger-toggle Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(frontend): make read-only toggle discreet in both modes Match the MCP-mode treatment in standard mode: text-tertiary, 2xs, shared "Read-only" label. The tooltip switches per mode so the explanation still fits the context. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(frontend): gate read-only toggle behind Limit token permissions The read-only toggle now only shows when the user has limited the token's scopes (standard mode) or in MCP mode (which always picks an MCP scope). Turning the limit off also resets read-only so it doesn't silently stick. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(frontend): hide incompatible MCP tools when read-only is on When the read-only toggle is on in MCP mode: - Endpoint badges and the custom-mode endpoint MultiSelect filter to GET. - Already-selected non-GET endpoints are pruned from the scope. - The scripts/flows preview is replaced with a note explaining they're hidden (the runner already rejects script/flow runs for read-only). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(frontend): place read-only toggle at top of limited scope area The previous gate required at least one scope to be picked before the read-only toggle appeared, which made it look missing while the user was still building their scope list. Move the toggle inside ScopesPicker: - Standard mode: sits directly under the "Limit token permissions" toggle whenever Limit is on, before the scope selector. - MCP mode: sits at the top of the MCP scope block. readOnly is now $bindable on ScopesPicker so CreateToken still owns the value. The auto-reset on un-limit moves into ScopesPicker too. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(frontend): nest read-only toggle inside the scope list card Place the read-only toggle at the top of the scope list (between the Selected Scopes summary and the bordered domain list) via a new optional topSlot snippet on ScopeSelector. Keeps ScopeSelector decoupled from read-only specifics; ScopesPicker fills the slot. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * chore: update ee-repo-ref to 9bc8160be50b3e57a60daf4e1b71c389a6e02b8a This commit updates the EE repository reference after PR #571 was merged in windmill-ee-private. Previous ee-repo-ref: f53d26e6685dfd60bfa67686fbd7358169cfd130 New ee-repo-ref: 9bc8160be50b3e57a60daf4e1b71c389a6e02b8a Automated by sync-ee-ref workflow. * fix: address CI review for read-only token flag - P1 (Codex): narrow the MCP middleware exemption from "any /api/mcp/*" to just the streamable HTTP transport endpoints (/api/mcp/gateway, /api/mcp/w/{ws}/{mcp,sse,list_tools}). Without this, a read-only token could POST /api/mcp/gateway/oauth/server/approve and mint a follow-on non-read-only MCP token via the OAuth code/token exchange. - P2 (Claude/cubic): fix test comment/assertion mismatch — the run-path assertion now exercises GET (which is what the RUN_PATH_ACTIONS elevation comment describes) in addition to POST. Add a regression assertion for /api/mcp/gateway/oauth/server/approve. - P2 (cubic): short-circuit script/flow/hub-script/resource fetches in MCP list_tools when read_only is on — they would only be discarded below, so skipping the DB and resource fan-out is pure win. - P2 (cubic): when scopes are pre-supplied via the CreateToken prop, the ScopesPicker isn't rendered, which previously hid the read-only toggle entirely. Render it next to the pre-supplied scopes display. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com> |
||
|
Ruben FiszelandClaude Opus 4.7
|
4d0f2c26a1 |
fix(bun): pass --preserve-symlinks on unbundled execution (#9147)
* fix(bun): pass --preserve-symlinks on unbundled execution Bun 1.2/1.3 moved its global package cache to a content-addressed layout and the installer now creates a single directory symlink from node_modules/<pkg> to the cache entry. Without --preserve-symlinks, Bun resolves modules from each file's realpath, so any require/import inside an installed package walks up from cache_nomount/bun/... and never finds the sibling deps living under <job_dir>/node_modules/. This manifested as e.g. ENOENT while resolving package 'zod/v3' from '/tmp/windmill/cache_nomount/bun/@langchain/core@1.1.44@@@1/dist/...' on //nobundling scripts that pull @langchain/core, even though zod is correctly installed alongside it in node_modules. The bundled execution path already had --preserve-symlinks since #4132 (needed because we symlink the cached bundle file into the job dir). The unbundled path didn't, because at the time Bun installed via per- file hardlinks and the realpath of node_modules entries was the job dir itself. The Bun installer's layout change made the flag necessary on the unbundled path as well. Add the flag to all three unbundled `bun run` invocations: - nsjail unbundled path - non-nsjail unbundled path - dedicated worker (always unbundled) This also fixes a latent bug on the first run of any bun script that imports a package whose internals reference siblings (the build_cache path runs unbundled this round while it builds the bundle for next time). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test(bun): regression test for nobundling + transitive require resolution Adds an integration test that mirrors the original failure: a //nobundling script importing @langchain/core, which (in its CJS internals) does require('zod/v3'). Before --preserve-symlinks was added to the unbundled bun run invocations, this failed with: ENOENT while resolving package 'zod/v3' from '.../cache_nomount/bun/@langchain/core@<ver>@@@1/dist/runnables/base.js' The test covers the non-nsjail unbundled path. Reproducibility of the pre-fix failure depends on Bun's installer choosing the directory-symlink layout for the node_modules entry (the default on Bun 1.2/1.3+ with the new content-addressed global cache that produced the user's error). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Ruben Fiszel
|
3a2901f6fe | update | ||
|
Diego ImbertandClaude Opus 4.6
|
4427a3d37f |
feat: add workspace-specific flag for resources and variables (#8836)
* feat: add workspace-specific flag for resources and variables Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: remove set_ws_specific endpoint and fix rust-client compilation Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: fall back to workspace name for ws_specific file naming When wsNameForFiles is not set (no wmill.yaml workspace config), ws_specific items would not get workspace-suffixed filenames during pull. Now falls back to workspace.name/workspaceId. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: use workspace ID instead of CLI name for ws_specific file naming Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: pass workspace ID fallback to elementsToMap for ws_specific push Without this, workspace-specific files (e.g., a.admins.resource.yaml) were not recognized during push when no wmill.yaml or git branch was available, causing spurious deletions. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * ui nits * nit * Fix variable edit when only editing ws_specific * mark_linked_variables_ws_specific * Helper label * Support json format alongside yaml * Fix file naming push/pull asymetry & ws_specific orphans * Revert all CLI diffs * CLI now appends the remote ws_specific list to the local specificItems * UI for Env switcher * Refactor Resource/Variable editors to use dumb component * Refactor side effects * Editor works with multi workspaces * Fix can_save * Fix As JSON * nit * UI nits * list_ws_specific_versions as pl sql function to avoid round trips * UI Nits * Per-workspace version read-only check * fix: reset session context in list_ws_specific_versions to prevent RLS leakage The function calls set_session_context() in a loop. Although SET LOCAL is transaction-scoped (so settings revert at autocommit), defending against the function being invoked inside a longer outer transaction: - wrap the loop in a sub-block with EXCEPTION WHEN OTHERS that resets the session to a deny-default (windmill_user, empty session.* GUCs) before re-raising, - on the happy path, reset to the same deny-default at the end of the function. * feat: audit auto-marked ws_specific variables When a resource is saved as ws_specific, every variable referenced via $var: inside its value is auto-INSERTed into ws_specific. Previously this happened silently. Now: - mark_linked_variables_ws_specific takes the authed user, - the INSERT uses RETURNING path so we know exactly which variables were freshly flipped (not the ones already ws_specific), - each newly flipped variable gets a 'variables.set_ws_specific' audit entry pointing at the resource that triggered it. * perf: skip mark_linked_variables_ws_specific when nothing relevant changed update_resource was calling mark_linked_variables_ws_specific on every save when the resource was ws_specific, even on a description-only or label-only edit. Gate the call on `ns.value.is_some() || ns.ws_specific == Some(true)` so we only re-mark when the $var: refs could actually have changed or ws_specific was freshly enabled. * docs: explain asymmetric ws_specific toggle in resource tooltip Enabling the resource's 'Workspace specific' toggle silently marks every variable referenced via $var: inside the value as ws_specific, but disabling it does not un-mark those variables (they may be referenced by other resources). Surface this in the tooltip so users know what to expect. * fix: surface non-404 errors when fetching ws_specific items in CLI sync mergeWsSpecificFromServer was catching every error from listWsSpecific and logging it at debug. That's correct for old servers without the endpoint (404), but a 401/403/network failure would silently produce an incomplete sync. Now distinguish 404 (debug, expected) from everything else (warn with status + message) so users notice when the merge fails for real reasons. * perf: collapse compare_two_variables presence checks into one round-trip The early-return path was issuing four sequential EXISTS queries (ws_specific × {source, fork}, variable × {source, fork}). Combine them into a single SELECT so the per-variable diff cost drops ~4x. * sqlx prepare * docs: clarify has_sql_updates invariant in update_variable The else branch of the npath resolution is only reachable for non-rename edits (labels-only, ws_specific-only) because ns.path being Some always forces has_sql_updates=true at the top of the function. Add a debug_assert and a comment explaining the invariant so a future change that decouples ns.path from has_sql_updates trips immediately. Also use `path` directly instead of unwrap_or_default-ing ns.path, since we know it's None here. * chore: drop redundant ws_specific type augmentations ListableResource and ListableVariable from $lib/gen now include `ws_specific?: boolean` after the openapi.yaml additions in this branch were regenerated. The intersection types in resources/+page and variables/+page were duplicating the field — drop them. * Put WsSpecificVersions toggle in top drawer bar * nit size * feat: detect local-only ws_specific items on sync push When wmill.yaml lists a resource/variable in specificItems but the remote isn't yet marked ws_specific for that item, sync push silently dropped the flag because: 1. file-content diff alone never noticed (ws_specific is metadata, not YAML body) — push{Resource,Variable} were never called for those items; 2. even when called, isSuperset(local, remote) returned true and the early-return skipped the API call. Now: - mergeWsSpecificFromServer returns the raw server list alongside the merged config so push can compare 'in local' vs 'in server'; - a new computeWsSpecificFlagOnlyPushes helper walks the local file map, finds ws_specific-flagged paths absent from the server list, and the push function injects them as synthetic 'edited' changes (same before and after content) so the standard display + apply pipeline picks them up; - push{Resource,Variable} no longer early-return when content matches but the ws_specific flag differs. Pull is unaffected — only the push-side caller of mergeWsSpecific takes the new (merged, serverItems) tuple. * getDeployTo for selected ws * refactor: ws_specific kind handling, support .json files The ws_specific helpers had two warts: 1. computeWsSpecificFlagOnlyPushes hardcoded `.resource.{yaml,json}` / `.variable.{yaml,json}` magic strings, even though the existing getTypeStrFromPath / removeType helpers already do that work and already cover both extensions. 2. isSpecificItem / isItemTypeConfigured only matched `.yaml` paths, so users with opts.json local files got no specificItems coverage at all — patterns from wmill.yaml (and from mergeWsSpecificFromServer) are expressed with `.yaml`, and a `.json` file never matched. Changes: - Replace WS_SPECIFIC_KIND_MAP (a closed enum of resource+variable) with configKeyForItemKind, a generic kind→SpecificItemsConfig key mapping. Triggers fold into 'triggers' via the `_trigger` suffix, so adding a kind to the backend's list_ws_specific_versions doesn't require a CLI change. - mergeWsSpecificFromServer now appends `${item.path}.${item.item_kind}.yaml` through the same helper. - computeWsSpecificFlagOnlyPushes uses getTypeStrFromPath + removeType, gated by configKeyForItemKind. No more magic strings. - isSpecificItem and isItemTypeConfigured normalize trailing `.json` to `.yaml` once at the entry, so a single set of patterns covers both extensions for the same logical item. * refactor: dedicated change type for ws_specific flag-only pushes Previously the sync push code injected a synthetic 'edited' Change with before === after to nudge the apply loop into calling pushResource / pushVariable for ws_specific-flag-only diffs, and a guard inside those two functions skipped the early-return when the flag differed. The contract was implicit and easy to break — any future 'skip identical edits' optimization in the change pipeline would silently drop these pushes. Replace with an explicit Change variant: type WsSpecificFlag = { name: 'ws_specific_flag'; path: string; kind: string; wsSpecific: boolean; }; The push apply loop now has a dedicated branch for it that calls wmill.updateResource / updateVariable with just the ws_specific flag. prettyChanges renders it on its own line. The dry-run JSON output picks it up via the existing change.name / change.path passthrough. The defensive wsSpecificMatches check inside push{Resource,Variable} is no longer needed (sync push doesn't go through them for flag-only diffs) and is reverted. * drop folders * feat(cli): warn on remote ws_specific items missing from local config When 'wmill sync pull' fetches the server's ws_specific list, items the server marks as ws_specific but that aren't matched by the local wmill.yaml's specificItems patterns now produce a warning. The merge already preserves correctness (those items are still treated as ws_specific during this pull), but the user's config drifts from the remote — and a later push from another machine without that config would push the item as non-ws_specific. Surface the drift so the user can update wmill.yaml. Also filter ws_specific_flag changes out before preCheckPermissionedAs (it expects added/edited/deleted only and they have no content payload so on_behalf_of resolution doesn't apply). * fix(cli): scope ws_specific drift warning to items in this pull's changes Previously the warning iterated every ws_specific item the server returned, producing log spam for items unrelated to the current pull (items that exist locally with no change, or items the user has nothing to do with this round). Move the loop after compareDynFSElement and only warn for items whose path appears in the changes list — i.e., items the user is actually pulling right now. * fix: clean up linked-side ws_specific rows on resource/variable delete Three places left orphaned ws_specific rows behind: 1. delete_resource deleted the resource's own ws_specific row and the linked variables, but never the ws_specific 'variable' rows that mark_linked_variables_ws_specific had auto-inserted for those variable paths. 2. delete_variable deleted its own ws_specific row and the linked resource at the same path, but never a ws_specific 'resource' row at that path. 3. delete_resources_bulk didn't even cascade to linked variables, let alone clean up their ws_specific rows. A new resource or variable later created at one of those paths would silently inherit a stale ws_specific flag — list_ws_specific would report it as workspace-specific, workspace diffs would treat it as 'no changes', and CLI sync would skip it. Fix: - delete_resource: DELETE FROM ws_specific WHERE item_kind = 'variable' AND path = ANY(linked_var_paths) before the linked-variable delete. - delete_variable: DELETE FROM ws_specific WHERE item_kind = 'resource' AND path = path before the linked-resource delete. - delete_resources_bulk: collect $var: refs from each bulk-deleted resource (mirror of single delete), then delete ws_specific 'variable' rows AND the variable rows themselves. Brings bulk delete in line with single delete semantics, including the orphan cleanup. * fix: gate list_ws_specific by resource/variable RLS The endpoint queried ws_specific directly under user_db, but ws_specific itself has no per-item RLS — only a workspace-level column. Any workspace member could enumerate every ws_specific path including those in folders they lack read access to (e.g. f/finance/prod_db_creds), revealing path existence that list_resources / list_variables would have hidden. Add EXISTS clauses against resource and variable so the same path-based RLS policies that govern those tables (see_own / see_member / see_extra_perms_user / see_extra_perms_groups / see_folder_extra_perms_user) also gate visibility here. The user transaction already establishes the session context; the joins make the policies apply. * only resources and variables * fix(cli): make workspace-specific path mapping handle .json files isSpecificItem() was extended to normalize .json -> .yaml so .json files could be matched against patterns, but the surrounding helpers remained yaml-only: - toWorkspaceSpecificPath only mapped folder.meta.yaml / settings.yaml / .X.yaml — a foo.resource.json went through unchanged, so the workspace-specific filename was never produced. - fromWorkspaceSpecificPath only matched .yaml extensions — pushing foo.dev.resource.json could not map back to foo.resource.json. - isCurrentWorkspaceFile / isWorkspaceSpecificFile regexes ended in \.yaml$, missing every branch-specific .json file. Replace the literal '.yaml' anchors with '(yaml|json)' alternations, preserve the actual extension on round-trips, and rename the helper buildYamlTypePattern -> buildItemTypePattern (it never had anything extension-specific in it). getFileTypeSuffix now returns the matching suffix for either extension. Changed: - getFileTypeSuffix - toWorkspaceSpecificPath / fromWorkspaceSpecificPath - isCurrentWorkspaceFile / isWorkspaceSpecificFile - isTriggerFile / isScheduleFile isItemTypeConfigured / isSpecificItem don't need touching — their checks run after normalizeJsonToYaml(), which already collapses both extensions to .yaml at the entry. * fix: create_resource?update_if_exists=true honors ws_specific=false The upsert path matched on `unwrap_or(false)`, so an explicit `ws_specific: false` and an absent flag were indistinguishable — both fell through with no DELETE on the existing ws_specific row. Callers trying to clear the flag via PUT-with-update_if_exists silently saw their request ignored. Mirror update_resource's three-way handling: Some(true) -> INSERT (+ mark linked variables) Some(false) -> DELETE (only when update_if_exists, since a pure create has no existing row anyway) None -> leave the existing flag alone create_variable doesn't have an upsert path (no ON CONFLICT), so the same bug doesn't apply there. * sqlx prepare * test: cover ws_specific cleanup, RLS filtering, upsert clearing, and CLI .json paths Backend (backend/tests/ws_specific.rs + fixture): - test_linked_delete_cleanup: creates a ws_specific resource that references a variable via $var:, deletes the resource, asserts the cross-kind ws_specific row for the auto-marked variable is also removed. Then does the inverse for delete_variable, verifying the ws_specific 'resource' row at the same path is cleaned by variable delete. - test_list_ws_specific_filters_by_rls: admin creates ws_specific items in u/test-user/ and u/test-user-2/; verifies admin sees both via list_ws_specific while a non-admin (test-user-2) only sees their own path — the RLS see_own policy on the joined resource/variable tables hides the other. - test_create_resource_upsert_clears_ws_specific: covers the three-way Option<bool> handling on the upsert path: Some(true) inserts, Some(false) clears the existing row, None leaves it alone. CLI: - specific_items_unit.test.ts: add 14 tests covering toWorkspaceSpecificPath / fromWorkspaceSpecificPath / isWorkspaceSpecificFile / isCurrentWorkspaceFile / isSpecificItem / isItemTypeConfigured for .json files (variable, resource, trigger, schedule, folder.meta, settings). - ws_specific_flag_only_unit.test.ts (new): covers computeWsSpecificFlagOnlyPushes — emits flag-only changes only for resource and variable kinds (the backend's list_ws_specific_versions scope), does not emit for schedules or triggers, returns empty when serverItems is null (older server), respects existing server entries, preserves .json extension on filePath. - Export computeWsSpecificFlagOnlyPushes so it can be unit-tested. * perf: index workspace_settings.deploy_to for the recursive CTE list_ws_specific_versions's recursive CTE probes WHERE ws.deploy_to = r.ws_id every iteration; without an index on workspace_settings.deploy_to each iteration seq-scans the table — at 10M workspaces with the depth cap of 32 that's up to 320M row reads per call. deploy_to is sparse (most workspaces don't deploy anywhere), so a partial index WHERE deploy_to IS NOT NULL stays small while still covering every probe. Tucked into the existing migration since the function and the index ship together. --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> |
||
|
|
e74f06cb56 |
fix: handle singlestepflow zombies and stop filtering them from runs page (#9055)
* fix: handle singlestepflow zombies and stop filtering them from runs page * fix: support singlestepflow in batch_rerun_jobs Previous PR added singlestepflow to list_selected_job_groups so the BatchReRun pane shows them, but batch_rerun_jobs_inner still joined on kind = 'script' / 'flow' with j.runnable_id (which is NULL for SingleStepFlow), so the rows were silently filtered out — user sees the option, click Re-run, gets zero successes. Mirror the norm_kind CTE projection from list_selected_job_groups inside batch_rerun_jobs_inner: pull the wrapped runnable type and pinned script hash from raw_flow.modules[id='a'], cast back to JOB_KIND so the existing handler dispatch works unchanged. Path-based schema fallback so input_transforms still resolve at rerun time. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: project singlestepflow in batch-rerun schema lookups Codex review pointed out two follow-on regressions from the previous fix: (1) list_selected_job_groups returned schemas with script_hash=null and schema=null for singlestepflow rows because the inner schemas subquery still joined runnable metadata via j.runnable_id (NULL for SingleStepFlow). The BatchReRun pane consumes every selected.schemas entry through mergeSchemasForBatchReruns / buildExtraLibForBatchReruns, both of which assume real schema objects. (2) When use_latest_version=true, batch_rerun_handle_job re-fetched latest_schema from v2_job filtering jb.kind='script' or 'flow' — neither matched singlestepflow, so schema came back NULL and every input_transforms entry silently no-op'd. Both queries now project singlestepflow rows via raw_flow.modules[id='a'] — norm_kind for dispatch and effective_hash for the schemas join, plus a path-based latest-schema fallback so flow-wrapped SSF (no version pinning) and any SSF whose pinned hash has been deleted still resolve. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test: add batch_rerun integration tests, fix SSF hash hex parsing Adds 11 integration tests against /jobs/run/batch_rerun_jobs and /jobs/list_selected_job_groups (both endpoints had zero CI coverage). Tests cover the full 4-kind × 3-mode matrix: regular Script and Flow (baseline regression for the SQL refactor), script-wrapped and flow- wrapped SingleStepFlow (regression for the bugs this PR fixes), and a mixed-kind batch. Writing the tests caught a real bug in the previous commit: ScriptHash serializes as a 16-char hex string in raw_flow.modules[a].value.hash (per the custom Serialize impl in windmill-types/scripts.rs), not as an integer. The earlier `(m->'value'->>'hash')::bigint` cast worked on the hand-inserted SQL fixture I'd used for live testing (which embedded the hash as a raw integer) but failed in production where all SSF jobs are pushed via JobPayload::SingleStepFlow's serialized form. Replaced with `('x' || lpad(hex, 16, '0'))::bit(64)::bigint` — preserves the twos-complement bit pattern so both positive and negative i64 hashes round-trip correctly. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * Update SQLx metadata --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com> |
||
|
Ruben Fiszel
|
5bca03eacd |
fix: bubble handle_flow chaining errors to parent flow (#9058)
* fix: bubble handle_flow chaining errors to parent flow * test: while-loop propagates inner forloop iterator failure |
||
|
Ruben Fiszel
|
eebaab9c87 | fix(bun): propagate non-zero exit from generate_bun_bundle on no-DB path (#9051) | ||
|
hugocasaandClaude Opus 4.7
|
6e5a21a9c7 |
fix(flows): inherit flow_env in sub-flow predicates (#9042)
* fix(flows): inherit flow_env in sub-flow predicates Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * refactor(flows): align flow_env lookup with get_root_job_id and tighten gate Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * refactor(flows): drop recursive CTE, root_job propagation suffices Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(flows): walk via flow_innermost_root_job to respect imported-flow scope Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * refactor(flows): remove flow_env API endpoint, dead code from deno_core era Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |