E
Alexandru Daicu (9f04de7f) at 17 Nov 09:45 test: [DEBUG] Log the background's filter lists in the restart test...
E
Vitalie Maldur (29b7ce0c) at 17 Nov 09:24 test: Save the Docker baselines of the host extensions again [FO-7]
E
Alexandru Daicu (cd1e6eed) at 17 Nov 09:23 test: [DEBUG] Log the settings page's reconnect in the restart test...
E
Alexandru Daicu (74bec1fe) at 17 Nov 08:36 test: Let the settings page start the service worker in the restart... ... and 11 more commits
E
Summary Fixes :-abp-contains() (and :has-text()) skipping elements nested inside an element that doesn't match, without slowing down filters that can't be affected. ContainsSelector.getElements() skipped every descendant of an element it had tested, even when that element's text didn't match. For most patterns that's safe: if a nested element's text matches, its ancestors' textContent contains the same text unbroken, so they match too. It's not safe for regular expressions that depend on the…
E
Wladislaw Schulmeister (419212e0) at 17 Nov 07:04 fix: remove reverted release notes ... and 4 more commits
E
Context We plan to collect which network filters matched on which site. Before writing that code, we need a reproducible way to show it does not make the extension worse. This MR adds only the test harness and changes no SDK source. It gives baseline numbers for today's build and is what the later changes will be measured with. It is a draft until the pipeline has run once, because the time the new test adds to the performance job on CI runners is not measured yet. Description New filter…
E
Context We plan to collect which network filters matched on which site. Before writing that code, we need a reproducible way to show it does not make the extension's service worker slower or heavier. Lighthouse on live sites cannot show that, it measures the page and varies a lot between runs. This is the second of two MRs split from !1626 (closed) and it is stacked on !1635. It changes no SDK source, only the test setup. It is a draft until !1635 is merged and the pipeline has run once,…
E
Vitalie Maldur (1e091d1c) at 16 Nov 22:12 test: Measure the service worker on local pages in the performance ... ... and 1 more commit
E
Context We plan to collect which network filters matched on which site. Before writing that code, we need a way to see what a change costs in AdBlock and Adblock Plus, not only in the SDK's test extension. This is the first of two MRs split from !1626 (closed). It only adds the host extensions to the lighthouse tests we already have, with the baseline we already have. Description New --extension={performance|adblock|adblockplus} option for test:performance. The default is the SDK's test…
E
Vitalie Maldur (2b1f3d4e) at 16 Nov 21:42 test: Run the lighthouse performance tests against the host extensi...
E
Vitalie Maldur (70e5775a) at 16 Nov 20:38 test: Add the Docker baselines of the metrics tests and tighten the...
E
Vitalie Maldur (c55f9246) at 16 Nov 16:57 test: Run the performance tests against the host extensions and mea...
E
Summary Implements LOVEAB-118. Stacked on !1620 (merged), so it targets feat/LOVEAB-72-language-list-pickers. New installs get EasyList plus language-only lists (!1613 (merged)). If the user removes or turns off EasyList, the language-only lists only block ads on websites in their language, and without a language list nothing blocks most ads. Adblock Plus doesn't say so. AdBlock has a notice for this (#easylist_info). This MR adds one to Adblock Plus's settings page, which answers LOVEAB-72's…
E
Alexandru Mihai (50dd9bab) at 16 Nov 16:13 feat: Show a notice while only language-only lists are enabled with...
E
Thomas Greiner (e183b3ac) at 16 Nov 16:03 perf: Emit host Prefs change events directly instead of via storage... ... and 2 more commits
E
Wladislaw Schulmeister (ff01559a) at 16 Nov 15:52 fix: remove reverted release notes ... and 4 more commits
E
Ursa Kacar (0171618f) at 16 Nov 15:50 refactor: Move the admin install check into a managed-installation ... ... and 2 more commits
E
Summary Implements LOVEAB-118. Stacked on !1620 (merged), so it targets feat/LOVEAB-72-language-list-pickers. New installs get EasyList plus language-only lists (!1613 (merged)). If the user removes or turns off EasyList, the language-only lists only block ads on websites in their language, and without a language list nothing blocks most ads. Adblock Plus doesn't say so. AdBlock has a notice for this (#easylist_info). This MR adds one to Adblock Plus's settings page, which answers LOVEAB-72's…
E
Alexandru Mihai (614cdc22) at 12 Nov 09:51 chore: Remove the unused AdBlock language filter list translations ...
E
You noted that this would have a performance impact for these filters, since they would be checking on every element. Did you do any testing of real sites with this change to check if there were significant performance impacts? As a follow-up suggestion, since regex anchors are the problem, maybe this change could only apply to filters that use a regex with anchors?
E
Context Part of LOVEAB-100: removing AdBlock's betafish-subscriptions.json and identifying filter lists the same way as Adblock Plus. Since !1617 (merged), the language filter lists are labelled with the native names of their languages, so their old translations are no longer used. Independent of !1619. Description Removed 92 unused translation keys from all locales: The 39 labels of the language filter lists that !1617 (merged) stopped using, e.g. filtereasylist_plus_german and…
E
Alexandru Mihai (4ea64cec) at 12 Nov 09:02 chore: Remove the unused AdBlock language filter list translations ...
E
Hi @jhonnythrough, could you please review this small change?
E
Summary Blocks eyeo/filters/ab-testing-config!255 Adds a blockedPerDay fact for the block count cohorts in LOVEAB-73. The SDK computes it, so AdBlock and Adblock Plus both get it without changes of their own. blockedPerDay is blockedCount ÷ (UTC days since installDate + 1), rounded down, so on the install day it equals the total so far. The SDK derives it in getAllFacts from those two host facts (core/sdk/src/background/remote-config/blocked-per-day.js). It's left undefined when the install…
E
I tested this branch against a local remote config server and I can see the block count and cohort assignment seems to be working correctly. Would be great if @ursakacar can review / test here too