~95% on SimpleQA (e.g. Qwen3.6-27B on a 3090). Supports all local and cloud LLMs (llama.cpp, Ollama, Google, ...). 10+ search engines - arXiv, PubMed, your private documents. Everything Local & Encrypted.
Python academiaanthropicarxivbravedeep-researchencryptionhome-automationhomeserverlocallocal-deep-researchlocal-llmmistralollamaopenaipubmedresearchresearch-toolretrieval-augmented-generationsearxngself-hosted
fix(tests): wait for async content to settle in mobile-safari nav tests (#4439)
LearningCircuite87da6b8e72d
diff --git a/tests/ui_tests/playwright/tests/all-pages-mobile.spec.js b/tests/ui_tests/playwright/tests/all-pages-mobile.spec.jsindex e21549a05d..c7db0b2b8c 100644--- a/tests/ui_tests/playwright/tests/all-pages-mobile.spec.js+++ b/tests/ui_tests/playwright/tests/all-pages-mobile.spec.js@@ -388,6 +388,35 @@ test.describe('All Pages - Content Above Mobile Nav', () => {// `load`, leaving zero time for the DOM walk below.await page.waitForLoadState('load', { timeout: 3000 }).catch(() => {});+ // Wait for async-loaded content to settle before scrolling.+ // History, News, Library and Metrics populate their lists via XHR+ // *after* `load` fires. History/News/Library have no content-ready wait+ // in waitForPageLoad at all, and Metrics only waits for its cards to+ // exist (not for the layout to stop shifting). A single scroll-to-bottom+ // taken mid-load lands on a stale layout; the list then resolves to its+ // empty-state and the page's bottom controls (search inputs, filter+ // dropdowns) end up behind the fixed mobile nav — which the assertion+ // below flags as "hidden behind nav". The #4314 clearance padding is+ // already sufficient; the failures were a measurement race, confirmed by+ // trace snapshots showing loading spinners still present at assertion time+ // on /history/, /news/, /library/, /metrics/.+ //+ // Poll until the document height is stable across two consecutive samples.+ // Bounded so pages with continuous lazy-loading (e.g. `/`) still proceed+ // instead of hanging — the exact failure mode that retired the old+ // scrollY-based poll referenced below.+ await page+ .waitForFunction(+ () => {+ const h = document.documentElement.scrollHeight;+ const prev = window.__ldrPrevScrollH;+ window.__ldrPrevScrollH = h;+ return prev !== undefined && Math.abs(h - prev) < 2;+ },+ { timeout: 4000, polling: 250 }+ )+ .catch(() => {});+// Scroll to bottom to check last content. scrollTo() is synchronous inside// page.evaluate(): the browser commits the scroll before the evaluate resolves,// so an additional `waitForFunction(scrollY >= scrollHeight)` poll is redundant.diff --git a/tests/ui_tests/playwright/tests/interactive-states.spec.js b/tests/ui_tests/playwright/tests/interactive-states.spec.jsindex ffb37e0083..452caacb4e 100644--- a/tests/ui_tests/playwright/tests/interactive-states.spec.js+++ b/tests/ui_tests/playwright/tests/interactive-states.spec.js@@ -217,9 +217,13 @@ test.describe('Mobile More Menu - Navigation', () => {if (await metricsItem.count() > 0 && await metricsItem.first().isVisible()) {await metricsItem.first().click();- await page.waitForLoadState('domcontentloaded');- // Should now be on the metrics page+ // Wait for the navigation to actually land on /metrics before+ // asserting. waitForLoadState('domcontentloaded') resolves immediately+ // when the current document is already loaded, so the URL check could+ // race the (JS-driven) navigation and still observe '/'. waitForURL+ // blocks until the address bar reflects the new page.+ await page.waitForURL(/\/metrics/, { timeout: 10000 });expect(page.url()).toContain('/metrics');}}