The For You (personalized) and Trending tabs ranked strictly by tier/score and
view count, which stacked a creator's whole back-catalogue into one block.
Add spreadAuthors() to the feed sorters: a gentle diversity re-rank that keeps
the caller's ranked order as the priority but avoids runs of the same author —
at each step it takes the highest-ranked remaining item whose author wasn't
among the last `window` picks (default 1), falling back to the plain top item
when every remaining candidate is a recent author. The #1 ranked item is always
chosen first, so "best-fit"/"most-viewed" stays topmost, and each author's items
keep their relative ranked order.
Applied to createForYouScoreSorter (personalized branch) and createTrendingSorter
(live items). Unlike interleaveByAuthor (full round-robin, used only for For
You's no-signal fallback + Explore), this preserves ranking rather than
flattening it, so each tab keeps its unique ordering process — just de-clustered.
Recent (chronological) and Explore (already diversified) are unchanged.
Tests: tests/feed-engine/spread-authors.test.mjs (7 — top preserved, no
avoidable adjacent dupes, per-author order kept, single-author/short/anon edge
cases, window:2 spread).
Co-Authored-By: Claude Opus 4.8 <[email protected]>
New FEED_TYPES.TRENDING tab: the active (recently-added) source ranked by view
count (recency tiebreak, trusted-muted last), gated by FEATURE_TRENDING_FEED
(default on). Builds on the now-trustworthy view counts (#4).
- createTrendingSorter reads counts from runtime.getViewCount, which resolves
each video's pointer via the app's canonical deriveVideoPointerInfo against the
shared viewCounter cache (new getVideoViewCountSnapshot export) — the same
counts the grid cards already load.
- Live re-rank: viewCounter now emits a coalesced onViewCountsChanged signal;
while the Trending view is active it debounce-re-runs the feed (cheap re-rank
over the in-memory active set, no relay re-fetch), so the order settles into
true trending as counts stream in. Cold cache reads as recency, then improves.
- Pipeline extracted to js/feedEngine/trendingFeed.js (feedCoordinator is at its
size cap; bumped the cap for the thin load/refresh delegations).
- Wiring: FEED_TYPES + flag, feedCoordinator cases, app.js wrappers,
applicationBootstrap registration (hides the sidebar link when flag off),
viewManager init + js/trendingView.js, views/trending.html, sidebar link.
Tests: tests/trending-sorter.test.mjs.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Explore shared the full active set with Recent/For You and leaned on freshness,
so it read like a reshuffled Recently-added. Give it a discovery identity:
- Weights: drop freshness (0.25 -> 0.12) and bump novelty (0.30 -> 0.35) so
recency stops dominating and topic-novelty leads.
- Add a soft followed-author penalty (weight 0.3): the scorer now reads
subscriptionAuthors from the Explore runtime and deprioritizes creators you
already follow, so non-followed creators surface ("new to you"). A penalty,
not a filter, so the feed stays full if you follow lots of people.
- More variety: lower the diversity sorter's MMR lambda (0.7 -> 0.5).
Also compacted For You's subscriptionAuthors reader to a one-liner to stay
under feedCoordinator's size cap while adding the same reader to Explore.
Tests: explore follows-penalty scoring; existing explore-scorer/diversity +
feedCoordinator suites green.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
For You shared the full active set with Recent/Explore and only re-ranked by a
blended score, so when a user's signals were thin it collapsed toward freshness
and looked identical to Recently-added.
Give it a structural identity:
- The scorer tags each item forYouTier: 2 = followed author, 1 = interest/
watch-topic match, 0 = other.
- The sorter leads with tier (your people first) regardless of raw score, then
ranks within tier; trusted-muted still sinks last.
- When NOTHING is tiered (logged-out / no follows / no interests) it falls back
to an author-interleaved discovery order (new interleaveByAuthor) so the feed
never reads as a chronological clone of Recently-added.
Tests: tests/for-you-tiering.test.mjs (tier beats score, tiers stack, muted
last, no-signal interleave, scorer tier tagging). Existing for-you-scorer suite
still green.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Watch-history cards rendered "Untitled video" / "Unknown" for newly watched
videos. nostrClient.allEvents stores CONVERTED video objects
(convertEventToVideo), which carry tags/pubkey but no `kind` field — only raw
relay events do. resolveEventAddress() did Number(event.kind) → NaN and bailed
with "", so the address (cache-by-"a"-pointer) scan matched nothing. Every video
pointer (all are "a" addresses) was forced to a relay round-trip, and when that
returned nothing the card showed Unknown — even though the just-watched video
was sitting right there in allEvents.
Fix: default to the video kind (30078) when a cached object has no `kind`, so the
local cache scan resolves freshly-watched "a" pointers without a relay round-trip.
Adds tests/watch-history-hydration-cache.test.mjs: relays return nothing, so a
pass can only come from the cache scan (mutation-verified against the NaN-reject
regression). Exports createWatchHistoryHydrationStage for boundary testing.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
The For You feed was chronological, hashtag-interest-EXCLUSIVE, and ignored
follows entirely (docs/feed-algo-audit.md). Now it is inclusive + ranked:
- createForYouScorerStage scores each video: follows-boost (author you follow)
+ interest/watch-topic affinity (cosine vs your interests+watch-history vector)
+ freshness - disinterest penalty. Reuses Explore's vector/similarity helpers.
- createForYouScoreSorter ranks by that score (muted last, recency tiebreak).
- For You pipeline: interests now BOOST not exclude (enforceInterests:false);
scorer added; chronological sorter replaced. Never empty (falls back to
recency when you have no follows/interests/history).
- buildForYouFeedRuntime now feeds subscriptionAuthors + watchHistoryTagCounts.
Watched/blocked/muted are still removed (suppression/blacklist/moderation
stages unchanged). Cheat-resistant tests (older boosted items only win via the
boost). Grid renders end-to-end (~1.3s). Weight tuning is a follow-up.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Replaces sequential await loop with Promise.any to resolve video timestamps in parallel.
This significantly reduces latency when multiple resolvers are present, especially if the first one is slow.
- Uses Promise.any to take the first valid finite timestamp.
- Wraps resolvers to reject on invalid values.
- Preserves error logging for genuine resolver errors.
- Benchmarks showed improvement from ~100ms to ~11ms in simulated slow scenarios.
Co-authored-by: PR0M3TH3AN <[email protected]>
- Extracted `getVideoTags` to `js/feedEngine/utils.js` with `WeakMap` caching.
- Updated `js/feedEngine/stages.js` and `js/feedEngine/kidsScoring.js` to use the cached utility.
- Removed redundant `collectVideoTags` and `normalizeTagSetFromVideo` functions.
- Added `benchmarks/feed_tags.mjs` to measure performance (approx 25x speedup in tight loops).
- Updated `hasDisinterestedTag` in `utils.js` to use the new utility.
This reduces object allocation and repeated string normalization during feed processing.
Co-authored-by: PR0M3TH3AN <[email protected]>
- Introduced `NORMALIZED_ARRAY_MARKER` in `js/feedEngine/utils.js` to tag arrays that are already normalized.
- Updated `normalizeItems` in `js/feedEngine/engine.js` to skip O(N) verification if the input array is marked.
- Updated built-in stages and sorters to mark their output arrays using `markAsNormalized` when they safely construct new arrays from normalized inputs.
- This avoids redundant O(N) scans after each stage, improving pipeline performance by ~10% end-to-end and eliminating overhead for trusted stages.
Co-authored-by: PR0M3TH3AN <[email protected]>
- Introduce `videoTagCache` (WeakMap) in `js/feedEngine/kidsScoring.js`
- Cache result of `normalizeTagSetFromVideo` to avoid re-computation
- Improved benchmark performance by ~35%
Co-authored-by: PR0M3TH3AN <[email protected]>
- Remove redundant double signer readiness gate evaluation that added 5-10s
to the login path (authSessionCoordinator)
- Reduce decryption timeouts (10s→6s per scheme, 3s→1.5s retry delay,
8s→5s NIP-07 no-prompt, 45s→25s overall ensureLoaded) for faster scheme
fallback during list sync
- Fix storage tab unlock: remove overly strict capabilities check that
blocked lazy-injecting NIP-07 extensions, and add auto-unlock attempt
when profile modal opens with an active signer
- Fix subscription/hashtag decryption: always resolve signer for background
refreshes (was previously gated behind allowPermissionPrompt), and always
check window.nostr as fallback decryptor regardless of prompt mode
- Fix capabilities check in subscription decryption: use method existence
instead of dynamic capabilities getter that returns false during lazy
extension injection
- Wrap Promise.any decryptor calls in Promise.resolve().then() to catch
synchronous throws from NIP-07 adapter methods when extension modules
haven't loaded yet
- Make watch-history feed registration race-safe with pre-check and
try/catch for concurrent registration attempts
- Improve NIP-07 extension detection with DOMContentLoaded listener for
faster detection of extensions that inject during page load
https://claude.ai/code/session_0174XTB5ADJ9cMBSucqL9Wer
Hoisted the timestamp hook resolution out of the inner loop in `createChronologicalSorter`.
This avoids allocating an array of candidate functions for every comparison during sorting (O(N log N)).
Benchmark shows a ~27% speedup (38ms -> 27ms for 100k items).
Co-authored-by: PR0M3TH3AN <[email protected]>
- Implements an Inverted Index to reduce redundant similarity checks.
- Replaces O(N) array splice with O(1) Map deletion for candidate tracking.
- Reduces runtime from ~3.5s to ~0.9s for 5000 items (3.9x speedup).
Co-authored-by: PR0M3TH3AN <[email protected]>
- Remove undefined hashtagPreferencesPromise reference in handleAuthLogin
that caused the feed sync Promise.allSettled to silently fail, preventing
proper list/profile synchronization before grid refresh
- Add missing getModerationSettings and getActiveProfilePubkey dependency
injection to authSessionCoordinator, fixing crashes during logout
moderation reset and profile switch pubkey resolution
- Parallelize NIP-07 permission request with relay readiness instead of
gating list loading behind it; parallelize subscriptions and hashtag
preference loading (independent of each other, only blocks must sequence)
- Add native run() alias to feedEngine so callers don't rely solely on
the applicationBootstrap compatibility shim
- Remove redundant refreshVisibleModerationUi call from renderVideoList
that re-decorated every video immediately after decoration
- Avoid re-normalizing moderation settings on every getActiveModerationThresholds
call when settings are already an object
https://claude.ai/code/session_01FCosieUXjEn95KtUeyGemY
- Introduce `NORMALIZED_MARKER` Symbol to mark normalized items.
- Optimize `normalizeDto` to return existing object if marked.
- Optimize `normalizeItems` to scan for markers and return input array if all items are valid and marked, avoiding unnecessary array allocation and object re-creation.
- Preserve safety logic in `runFeed` (handle nulls and always call normalizer) while benefiting from internal optimization.
Benchmark results show ~9x speedup for "Same Instance" updates and ~5x speedup for "Copy Array" updates in the feed pipeline.
Co-authored-by: PR0M3TH3AN <[email protected]>
- Deleted `js/magnet.js` as it was a redundant wrapper for `js/magnetUtils.js`.
- Refactored `js/services/videoNotePayload.js`, `js/ui/components/EditModal.js`, and `js/nostr/nip71.js` to use `magnetUtils.js` and `magnetShared.js`.
- Updated `tests/magnet-utils.test.mjs` to remove tests for the deleted file.
- Removed unused exports `DEFAULT_FEED_CONFIG` and `DEFAULT_CONFIG_SCHEMA` from `js/feedEngine/engine.js` and `index.js`.
- Verified changes with unit tests.
- Implements `hasDisinterestedTag` helper in `js/feedEngine/utils.js` to check for disinterested tags in both raw Nostr tags and NIP-71 metadata.
- Updates `createBlacklistFilterStage` in `js/feedEngine/stages.js` to filter out videos matching `context.runtime.tagPreferences` disinterests.
- Ensures existing blacklist logic remains undisturbed.
- Verified with new unit test `tests/feed-engine-tag-pref.test.mjs` and existing feed engine tests.
This change introduces `createWatchHistoryProfileHydrationStage` to the
watch history feed pipeline. The stage resolves author profiles for
hydrated videos by:
1. Checking the global `getProfileCacheEntry` for cached profiles.
2. Batching missing profiles into a single Kind 0 relay request.
3. Updating the global cache (`setProfileCacheEntry`) with fetched data.
4. Attaching the resolved profile to `item.metadata.profile`.
This ensures that channel images populate correctly in the watch history
UI (both sidebar and profile modal).
This commit addresses issues with the watch history not updating dynamically and adds a manual refresh mechanism.
Changes:
- Added a "Refresh" button to the watch history view in `views/history.html` and `js/historyView.js`.
- Updated `watchHistoryService.js` to accept a `forceRefresh` option in `loadLatest`, which bypasses the cache freshness check and triggers a relay fetch.
- Updated `js/feedEngine/watchHistoryFeed.js` to propagate the `forceRefresh` flag from the runtime context to the service.
- Added user feedback (toast messages and logs) for "Clear Local History" and "Republish Now" actions.
- Updated `tests/watch-history.test.mjs` to align with the new `fetchHistory` signature.
- Enforce descending month order in `js/nostr/watchHistory.js` by explicitly sorting keys before flattening buckets.
- Update `js/feedEngine/watchHistoryFeed.js` and `js/historyView.js` to use `video.created_at` as a fallback sort key when `watchedAt` is missing or equal.
- Verified logic with `tests/verify_sort.mjs`.
The watch history hydration process was sometimes failing to retrieve video metadata, resulting in "Untitled video" cards. This was primarily due to only querying the generic relay list and potentially missing content hosted on specific relays.
This commit improves the robustness of `createWatchHistoryHydrationStage` in `js/feedEngine/watchHistoryFeed.js` by:
1. Collecting unique relay hints from the pointer objects (`item.pointer.relay`) of items needing hydration.
2. Merging these hints with the user's `readRelays` (or fallback `relays`) list. This ensures that if a specific video is known to exist on a niche relay, that relay is queried during the metadata fetch.
3. Replacing conditional debug logging with `devLogger.info` and `devLogger.warn` to provide detailed insights into the hydration process (item counts, selected relays, generated filters, and fetch results), aiding in future diagnosis.
These changes ensure the application attempts to fetch metadata from all relevant sources indicated by the history data itself.
The watch history hydration stage was failing silently when relay connections were unstable or misconfigured. This commit:
1. Replaces the conditional debug logger with `devLogger.info` and `devLogger.warn` to ensure visibility of critical hydration steps.
2. Adds detailed logging for item counts, missing ID/address lists, filter generation, and fetch results.
3. Updates the relay selection logic to prioritize `nostrClient.readRelays` over `nostrClient.relays`. This ensures that hydration attempts query the user's specific read-configured relays, which is crucial for fetching metadata when generic relays may be write-only or restricted.
These changes aid in diagnosing "Untitled video" issues and improve the robustness of the metadata fetch process.
The `createWatchHistoryHydrationStage` in the feed engine was incorrectly relying solely on `nostrClient.relays` for network fetches. This could cause hydration failures if `relays` contained write-only servers or was otherwise restricted. This commit updates the logic to prioritize `nostrClient.readRelays` if available, falling back to `nostrClient.relays`. This ensures that hydration attempts query the correct relays configured for reading data.
Fixes an issue where watch history items would display as "Untitled video" due to failed metadata resolution.
Updated `createWatchHistoryHydrationStage` in `js/feedEngine/watchHistoryFeed.js` to first check `nostrClient.allEvents` for video data before querying relays. This ensures that recently watched videos (which are already in memory) are instantly hydrated in the watch history list, fixing the "Untitled video" issue.
Additionally, implemented robust support for hydrating address pointers (`type: 'a'`) by scanning the cache and constructing proper relay filters for missing items. This covers cases where history items reference videos by `kind:pubkey:d-tag` rather than ID.
Updated `createWatchHistoryHydrationStage` in `js/feedEngine/watchHistoryFeed.js` to properly handle and hydrate watch history items that use address pointers (type 'a'), specifically fixing a bug where d-tags containing colons were being ignored.
Previously, only event ID pointers (type 'e') were being hydrated, and an initial fix for addresses failed for d-tags with colons. This change robustly parses address pointers, constructs appropriate filters, fetches the events, and correctly maps them back to the history items using a `normalizeAddressKey` utility that handles mixed-case pubkeys and complex d-tags.
Updated `createWatchHistoryHydrationStage` in `js/feedEngine/watchHistoryFeed.js` to properly handle and hydrate watch history items that use address pointers (type 'a').
Previously, only event ID pointers (type 'e') were being hydrated, causing videos watched via address pointers to display as "Untitled video" with missing metadata. This change parses the address pointers, constructs appropriate filters (normalizing pubkeys to lowercase while preserving d-tag case), fetches the events, and correctly maps them back to the history items using a robust `normalizeAddressKey` utility.
Updated `createWatchHistoryHydrationStage` in `js/feedEngine/watchHistoryFeed.js` to properly handle and hydrate watch history items that use address pointers (type 'a').
Previously, only event ID pointers (type 'e') were being hydrated, causing videos watched via address pointers to display as "Untitled video" with missing metadata. This change parses the address pointers, constructs appropriate filters, fetches the events, and correctly maps them back to the history items.
This introduces a new stage in the watch history feed pipeline that fetches missing video metadata from relays. Since watch history storage was optimized to store only event IDs, this hydration step ensures that the UI can display titles and thumbnails without requiring them to be persisted in the history payload itself.
- Implemented `createWatchHistoryHydrationStage` in `js/feedEngine/watchHistoryFeed.js`.
- Integrated `convertEventToVideo` to transform fetched events into usable video objects.
- Added a comprehensive integration test in `tests/watch-history.test.mjs` verifying the full flow from ID-only storage to hydrated feed items.
Removes the metadata argument from `WatchHistoryService.publishView` and cleans up the associated caching and resolution logic in `watchHistory.js` and `watchHistoryFeed.js`. Deletes `js/watchHistoryMetadata.js` as it is no longer used.
This simplification ensures watch history strictly manages pointers and timestamps, leaving metadata resolution to higher-level consumers. It also reduces the payload size and complexity of persisted history records.
Tests have been updated to reflect the API changes.
